fix: allow e2e enrollment feature without fixtures

fix: split rustfs-cli e2e entrypoint
refactor(object): extract internal put entry and ODM provenance keys (#7071 )
2026-09-02 23:53:24 +08:00 · 2026-09-02 23:26:07 +08:00 · 2026-09-02 22:56:23 +08:00 · 2026-09-02 22:54:51 +08:00 · 2026-09-02 22:46:28 +08:00 · 2026-09-02 14:31:29 +00:00
882 changed files with 191100 additions and 57416 deletions
@@ -50,10 +50,11 @@ consider adding it to the script's `checked_files` list.

 ## `check_doc_paths.sh`

-Instruction/architecture docs (`AGENTS.md`, `CLAUDE.md`, `ARCHITECTURE.md`,
-`docs/architecture/*.md`) must not reference repo file paths that no longer
-exist. If your refactor moved code, update the docs that point at it — the
-error message lists `doc -> stale-path` pairs.
+Instruction docs (`AGENTS.md`, `CLAUDE.md`, `ARCHITECTURE.md`) and every
+Markdown file under `docs/` (architecture, operations, testing, index) must not
+reference repo file paths that no longer exist. If your refactor moved code,
+update the docs that point at it — the error message lists `doc -> stale-path`
+pairs. Cite paths plus symbol names, never line numbers (see `docs/README.md`).

 ## `check_no_planning_docs.sh`

@@ -1,12 +1,12 @@
 ---
 name: rustfs-release-publish
-description: "Run the end-to-end RustFS console gate, version bump, preview validation, and final-tag publication pipeline. Use only when the user explicitly asks to release or publish a RustFS version (发版/发布)."
+description: "Run the end-to-end RustFS console gate, version bump, preview validation, human confirmation, and final-tag publication pipeline. Use only when the user explicitly asks to release or publish a RustFS version (发版/发布)."
 ---
 # RustFS Release Publish (preview-validated pipeline)

 This skill orchestrates a full release. It wraps `rustfs-release-version-bump` (which only edits version files and opens the PR) with a mandatory preview-tag validation loop before the final tag is published.

-Core design: **version files never carry a `-preview.N` suffix**. The preview suffix exists only in tag names. A preview tag creates a visible GitHub Release marked Prerelease and uploads versioned assets, but it never becomes GitHub Latest and never updates `*-latest`, `latest.json`, R2, Docker, or Helm channels. This works because the binary self-reports the git tag it was built from (`build::TAG` via shadow_rs, see `rustfs/src/config/cli.rs` `SHORT_VERSION`), and `build.yml` derives artifact names and preview classification from the tag name — Cargo.toml's version is only a no-tag fallback. Therefore the preview tag and the final tag can (and MUST) point at the exact same commit: what you validated is byte-for-byte the source that ships.
+Core design: **version files never carry a `-preview.N` suffix**. The preview suffix exists only in tag names. A preview tag creates a visible GitHub Release marked Prerelease and uploads versioned assets, but it never becomes GitHub Latest and never updates `*-latest`, `latest.json`, R2, Docker, or Helm channels. That Release is temporary: `build.yml` deletes it automatically once the final tag's Release is published, so the Releases page ends up carrying deliverables only while the `-preview.N` tags stay behind as the traceability record. This works because the binary self-reports the git tag it was built from (`build::TAG` via shadow_rs, see `rustfs/src/config/cli.rs` `SHORT_VERSION`), and `build.yml` derives artifact names and preview classification from the tag name — Cargo.toml's version is only a no-tag fallback. Therefore the preview tag and the final tag can (and MUST) point at the exact same commit: what you validated is byte-for-byte the source that ships.

 Pipeline shape:

@@ -17,7 +17,9 @@ check console main against its latest Release
  -> tag <preview-tag> at that commit -> CI green
  -> verify preview Release assets -> run binary locally + console checks
  -> validate with latest rc client
+  -> report preview acceptance results -> STOP for explicit human confirmation
  -> tag <target> at the SAME commit (zero delta) -> re-verify CI/release
+  -> CI deletes the <target>-preview.N Releases (tags kept)
 ```

 On validation failure: fix lands on main via normal PR (version files are already at `<target>`, no new bump PR), then tag `<preview-tag N+1>` at the new main commit and restart from Phase 2.
@@ -50,16 +52,20 @@ Rules:
 - Use `<target>-preview.N` for every target, e.g. `1.0.0-beta.10-preview.3` or `1.1.0-preview.1`.
 - The canonical suffix is exactly `-preview.<digits>`. `build.yml` recognizes it before alpha/beta/rc classification and routes it to the preview-only path; any other tag containing `-preview` fails closed instead of being treated as a release.
 - A preview Release MUST be published with `isPrerelease=true` and `isLatest=false`. Any `*-latest` preview asset or preview-triggered `latest.json`, R2, Docker, or Helm publication is a pipeline failure.
+- Preview Releases are cleaned up by the `cleanup-preview-releases` job after `publish-release` succeeds for the deliverable tag. It deletes every Release whose tag is exactly `<target>-preview.<digits>` and never passes `--cleanup-tag`, so the tags survive.

 ## Hard rules

 - Version files (Cargo.toml, Cargo.lock, README, flake.nix, Chart.yaml, rustfs.spec) are bumped ONCE, directly to `<target>`. Never write a `-preview.N` suffix into any version file. If `rustfs-release-version-bump` is ever asked for a `-preview` version, that is a pipeline bug — stop.
- Preview Release assets are versioned and intentionally visible on the Releases page. Do not label them Latest or use them to update any latest distribution channel.
+- Preview Release assets are versioned and intentionally visible on the Releases page for the duration of validation. Do not label them Latest or use them to update any latest distribution channel.
+- Never delete a preview Release by hand before Phase 6 finishes — Phase 4 downloads its assets and the final Release notes are generated while it still exists. Cleanup is CI's job; only step in manually (`gh release delete "<preview-tag>" --yes`, never `--cleanup-tag`) if `cleanup-preview-releases` failed.
 - Tags have no `v` prefix. Always annotated: `git tag -a <tag> -m "Release <tag>"`.
 - The final tag MUST point at exactly `PREVIEW_HASH` — the commit the validated preview tag points at. Never tag current `main` HEAD (commits merged after validation are unvalidated), and never create an extra version-bump commit between preview and final.
- When a previous deliverable exists, GitHub Release notes for the preview and final tags MUST use it as their shared comparison baseline: the most recently published non-preview Release before the target. Internal `-preview.N` Releases are explicitly excluded from that selection, even when they point at the same commit as the final tag. If no previous deliverable exists, omit `previous_tag_name` and record that GitHub's default baseline fallback was used.
+- When a previous deliverable exists, GitHub Release notes for the preview and final tags MUST use it as their shared comparison baseline: the most recently published non-preview Release before the target. Internal `-preview.N` Releases are explicitly excluded from that selection, even when they point at the same commit as the final tag — cleanup runs after the notes are generated, so the preview Release is still present and would otherwise be picked as the baseline. If no previous deliverable exists, omit `previous_tag_name` and record that GitHub's default baseline fallback was used.
 - Generated Release notes carry a workflow-management marker so retries can repair them. Before manually curating a generated body, remove that marker; unmarked non-placeholder notes are preserved by later workflow runs.
 - Phases run in order; a failure in any phase blocks everything after it. After the fix lands on main, restart from Phase 2 with the next preview iteration against the new `origin/main` hash — do not resume mid-pipeline against a stale hash.
+- Completing preview acceptance does not authorize the final tag. After Phases 3–5 pass, report the acceptance evidence and stop until the user explicitly confirms continuation. The original release request, an earlier confirmation, silence, or an automated follow-up does not satisfy this gate.
+- Confirmation is scoped to the reported `<target>`, `<preview-tag>`, and `PREVIEW_HASH`. A failed or repeated acceptance cycle, including any new preview iteration, invalidates prior confirmation and requires a new one.
 - If the release is abandoned after Phase 1 merged, main's version files claim a version that was never tagged. Either revert the bump PR or leave it to be overwritten by the next release — but tell the user explicitly and record the decision.
 - User-facing status updates in Chinese; commits, PR titles/bodies, and tag messages in English. No hard-wrapping in commit messages, PR bodies, or documentation prose — one logical line per sentence/paragraph, let soft wrap handle display.

@@ -207,6 +213,12 @@ rc alias remove preview

 - Any FAIL blocks the release. Afterwards stop the server and delete the scratch data directory.

+### Manual confirmation gate
+
+After every Phase 3–5 check passes, report the target, preview tag, `PREVIEW_HASH`, preview Release URL, console result, and rc matrix, then explicitly ask the user whether to publish the final tag. End the turn without creating or pushing `<target>`.
+
+Continue to Phase 6 only after a new user reply explicitly confirms the reported target, preview tag, and commit. A clear affirmative reply to that exact report, such as `确认继续`, is sufficient; if the reply is ambiguous or any reported value changed, ask again.
+
 ## Phase 6 — Publish the final tag on the validated commit

 No second version bump, no release branch. The final tag goes on the exact commit the preview validated:
@@ -221,6 +233,7 @@ git push origin "<target>"
 - CI rebuilds from the same source; the only changed input is the tag name, so the binary now self-reports `<target>`.
 - Verify the final tag's complete publication path: all matrix and release jobs green; `gh release view "<target>"` shows the full versioned and `-latest` asset set plus checksums, SBOM, and provenance; Docker and Helm workflows succeed; `latest.json` points to `<target>`. A stable target must have `isPrerelease=false` and `isLatest=true`. An alpha/beta/rc target must have `isPrerelease=true`; GitHub does not permit prereleases to be Latest, but the project `latest.json` still advances to the final non-preview target.
 - Verify the final Release body contains `## What's Changed` and a Full Changelog link. When `PREVIOUS_DELIVERABLE` exists, the link MUST be `https://github.com/rustfs/rustfs/compare/<PREVIOUS_DELIVERABLE>...<target>` and the baseline MUST equal the preview Release baseline; for example, both `1.0.0-beta.12-preview.1` and `1.0.0-beta.12` compare from `1.0.0-beta.11`.
+- Verify the preview cleanup: `cleanup-preview-releases` must succeed, `gh release view "<preview-tag>"` must then report `release not found` for every preview iteration of this target, and `git rev-parse "<preview-tag>^{commit}"` must still resolve to `PREVIEW_HASH` (the tag is kept). If the job failed, delete the leftover Releases manually with `gh release delete "<preview-tag>" --yes` and report it.
 - Optionally spot-check `./rustfs --version` from a final-tag artifact — it must report `<target>`.

 ## Output contract
@@ -229,5 +242,6 @@ Always report:

 - Console gate result: previous/latest Console tags, whether merged changes required a release, `CONSOLE_HASH`, and Console run/Release URLs when a release was published.
 - Target version, preview tag(s) used, `PREVIEW_HASH` (which both tags point at).
- Per-phase result (PASS/FAIL/BLOCKED) with key evidence: preview and final Release URLs, preview `isPrerelease`/`isLatest` state, final latest-channel state, console check results, and the rc command matrix.
+- Manual confirmation gate status (`WAITING_FOR_CONFIRMATION` or `CONFIRMED`) and its exact target, preview tag, and `PREVIEW_HASH`.
+- Per-phase result (PASS/FAIL/BLOCKED) with key evidence: preview and final Release URLs, preview `isPrerelease`/`isLatest` state, final latest-channel state, console check results, the rc command matrix, and the preview-Release cleanup result (deleted Releases plus surviving tags).
 - Any deviation from this pipeline and why the user approved it.
@@ -48,6 +48,7 @@ Update this file only when an advisory adds or changes a reusable lesson, affect

 ### S3 object actions, copy, multipart, and upload policy validation

+- `GHSA-g8w9-qw9q-fghr`: a valid presigned `PutObject` accepted extra `x-amz-tagging`, website redirect, and storage-class headers omitted from `SignedHeaders`. Lesson: a presigned URL is a bounded capability; reject `x-amz-*` headers that are not cryptographically bound by the signature so unsigned metadata cannot change authorization, lifecycle, redirect, cost, or durability semantics.
 - `GHSA-3ppv-fx5m-m749`: explicit `versionId` reads and copy sources authorized `s3:GetObject` instead of `s3:GetObjectVersion`. Lesson: version-specific object access must select version-specific actions for direct reads, `CopyObject`, and `UploadPartCopy`, with tests proving the backend is not reached on denial.
 - `GHSA-x298-9x87-fvjq`: anonymous `ListObjectVersions` fell back to `ListBucket` and returned before public-access-block gates. Lesson: compatibility fallbacks must converge on the same post-authorization checks as direct grants, especially `RestrictPublicBuckets` and anonymous data-plane denies.
 - `GHSA-mx42-j6wv-px98`: `UploadPartCopy` missed source authorization and allowed cross-bucket object exfiltration. Lesson: multipart copy must enforce the same source and destination contract as `CopyObject`.
@@ -119,7 +120,7 @@ Use these targeted searches when a diff touches security-sensitive code:
 ```bash
 rg -n "validate_admin_request|check_permissions|AdminAction::|deny_only|is_allowed" rustfs crates
 rg -n "authorize_operation|FtpsDriver|SftpDriver|RETR|MKD|SIZE|MDTM|CreateBucket|GetObject|HeadObject" crates/protocols rustfs
-rg -n "UploadPartCopy|upload_part_copy|CompleteMultipart|PostObject|content-length-range|starts-with" rustfs crates
+rg -n "UploadPartCopy|upload_part_copy|CompleteMultipart|PostObject|presign|SignedHeaders|content-length-range|starts-with" rustfs crates
 rg -n "ListBucketVersions|GetObjectVersion|versionId|VersionId|ExistingObjectTag|ForAllValues|ForAnyValue|POLICY_PLUGIN|opa" rustfs crates
 rg -n "normalize_extract_entry_key|Snowball|auto-extract|PathBuf::join|canonicalize|\\.\\.|x-forwarded-for|x-real-ip|SourceIp" rustfs crates
 rg -n "DEFAULT_SECRET|DEFAULT_ACCESS|TEST_PRIVATE_KEY|rustfs rpc|RUSTFS_RPC_SECRET" rustfs crates
@@ -136,6 +137,7 @@ rg -n "deny_unknown_fields|serde.default|as u32|as usize|as i32" rustfs crates
 - Protocol frontend authz fixes: include denied `RETR`, `SIZE`/`MDTM`, `MKD`, bucket probe, and sibling allowed-operation cases, and assert denied paths do not reach the storage backend.
 - IAM fixes: include import/update/list service-account cases with attacker-controlled parent, claims, access key, secret key, and policy.
 - Copy/upload fixes: include cross-bucket, cross-user, source-denied, destination-denied, copy-source-condition, and multipart completion cases.
+- Presigned upload fixes: include a valid presign with extra unsigned tagging, redirect, and storage-class headers; require rejection before storage access, and verify explicitly signed equivalents still work.
 - Version-action fixes: include historical UUID, explicit current version, `null`, range, partNumber, presigned, STS/session, service-account, anonymous bucket-policy, copy source, and multipart-copy source cases.
 - Policy-condition fixes: include reserved-key header collisions, missing keys, partially overlapping multi-value sets, plugin mode, and built-in policy mode.
 - Path fixes: include encoded traversal, absolute path, nested traversal, archive entries with `..`, valid object keys that resemble traversal text but should be rejected, and canonical bucket/prefix boundary checks.
@@ -1,2 +1,2 @@
-sha256-darwin=ae412c27e2e43f3fda3e0c21ef616b48a6403354faab8db8d4154202e3c4eec5
-sha256-linux=ef3be856bd3257c2c369428f66a48ad073dad8187229fe95840a600a81edf22b
+sha256-darwin=9dccb0cd537cf79ae70c1c20e8281d36d03f2f09f81142a5341e26e3dc18709d
+sha256-linux=a8a816d7bb0e7cb5632b1863b33794bcb9fc7e765f150aa5e1bf16518e28dfb4
@@ -1 +1 @@
-sha256=9b9bc336b43b70d0e06e0adb5455bf035bb18945d85d60936eb6fe4d48e0e680
+sha256=d06524b44de97ed8f62b0fd8cf9fa504e3cd520ffcaacc32691d6f890ebe7f20
@@ -1 +1 @@
-sha256=655a3f3c1d042e694339d15caba7580518320322d1bac0f09450b37e6c09e2e7
+sha256=8d5517f5f2fc32d561782dfccd51b7f746f5e25b2835e37e100c883f7f18777d
@@ -1 +1 @@
-sha256=ec27cde6ce6400723c4b372bfbd2ac61709c744294e4810af765e8a808d8e31d
+sha256=db9bd8cdcb0abe43461aa6b36499b17cabd4098e5b34e300b1a0f0d0f34d9884
@@ -23,4 +23,4 @@ coverage: core-deps ## Workspace line coverage (cargo-llvm-cov + nextest; slow,
 	@mkdir -p target/llvm-cov
 	cargo llvm-cov report --lcov --output-path target/llvm-cov/lcov.info
 	cargo llvm-cov report --json --output-path target/llvm-cov/coverage.json
-	python3 scripts/coverage_per_crate.py target/llvm-cov/coverage.json
+	$(RUSTFS_PYTHON_BIN) scripts/coverage_per_crate.py target/llvm-cov/coverage.json
@@ -45,6 +45,11 @@ logging-guardrails-check: ## Check logging guardrails for redaction and noise re
 	@echo "🪵 Checking logging guardrails..."
 	./scripts/check_logging_guardrails.sh

+.PHONY: error-other-ratchet-check
+error-other-ratchet-check: ## Check the ecstore ::other(format!) quorum-bucketing ratchet stays shrink-only
+	@echo "🪣 Checking error other(format!) ratchet..."
+	./scripts/check_error_other_format_ratchet.sh
+
 .PHONY: tokio-io-uring-check
 tokio-io-uring-check: ## Check tokio io-uring runtime feature stays removed
 	@echo "🚫 Checking tokio io-uring feature guard..."
@@ -75,10 +80,15 @@ embedded-secrets-check: ## Check no private key material or credential literal i
 	@echo "🔑 Checking embedded secret material guard..."
 	./scripts/check_embedded_secrets.sh

+.PHONY: offline-enrollment-e2e-check
+offline-enrollment-e2e-check: core-deps ## Build and exercise the dedicated offline enrollment E2E root
+	@echo "🔐 Checking the offline enrollment E2E root boundary..."
+	./scripts/check_offline_enrollment_e2e.sh
+
 .PHONY: test-wiring-check
 test-wiring-check: ## Check tests stay registered and selected by their intended runners
 	@echo "🧪 Checking test wiring..."
-	python3 ./scripts/check_test_wiring.py
+	$(RUSTFS_PYTHON_BIN) ./scripts/check_test_wiring.py

 .PHONY: log-analyzer-rules-check
 log-analyzer-rules-check: core-deps ## Check log-analyzer rule anchors still exist verbatim in source
@@ -19,13 +19,13 @@ planning-docs-check: ## Check that no planning-type documents are committed
 	./scripts/check_no_planning_docs.sh

 .PHONY: pre-commit
-pre-commit: fmt-check unsafe-code-check architecture-migration-check logging-guardrails-check tokio-io-uring-check extension-schema-check body-cache-whitelist-check s3s-footprint-check fips-wording-check embedded-secrets-check test-wiring-check doc-paths-check planning-docs-check quick-check ## Run fast pre-commit checks without clippy/full tests
+pre-commit: fmt-check unsafe-code-check architecture-migration-check logging-guardrails-check error-other-ratchet-check tokio-io-uring-check extension-schema-check body-cache-whitelist-check s3s-footprint-check fips-wording-check embedded-secrets-check test-wiring-check doc-paths-check planning-docs-check quick-check ## Run fast pre-commit checks without clippy/full tests
 	@echo "✅ All pre-commit checks passed!"

 .PHONY: pre-pr
-pre-pr: fmt-check unsafe-code-check architecture-migration-check logging-guardrails-check tokio-io-uring-check extension-schema-check body-cache-whitelist-check s3s-footprint-check fips-wording-check embedded-secrets-check test-wiring-check doc-paths-check planning-docs-check log-analyzer-rules-check clippy-check test ## Run full pre-PR checks with clippy and tests
+pre-pr: fmt-check unsafe-code-check architecture-migration-check logging-guardrails-check error-other-ratchet-check tokio-io-uring-check extension-schema-check body-cache-whitelist-check s3s-footprint-check fips-wording-check embedded-secrets-check test-wiring-check doc-paths-check planning-docs-check log-analyzer-rules-check offline-enrollment-e2e-check clippy-check test ## Run full pre-PR checks with clippy and tests
 	@echo "✅ All pre-PR checks passed!"

 .PHONY: dev-check
-dev-check: fmt-check unsafe-code-check architecture-migration-check logging-guardrails-check tokio-io-uring-check extension-schema-check body-cache-whitelist-check s3s-footprint-check fips-wording-check embedded-secrets-check test-wiring-check doc-paths-check planning-docs-check quick-check ## Run fast local development checks
+dev-check: fmt-check unsafe-code-check architecture-migration-check logging-guardrails-check error-other-ratchet-check tokio-io-uring-check extension-schema-check body-cache-whitelist-check s3s-footprint-check fips-wording-check embedded-secrets-check test-wiring-check doc-paths-check planning-docs-check quick-check ## Run fast local development checks
 	@echo "✅ Fast development checks passed!"
@@ -34,13 +34,15 @@ script-tests: ## Run shell script tests
 	./scripts/test_exact_1mib_handoff_abba.sh
 	./scripts/test_pinned_paired_abba_bench.sh
 	./scripts/test_manual_transition_runbooks.sh
+	./scripts/test_fuzz_runner.sh
+	./scripts/test_python_bin.sh
 	./scripts/check_embedded_secrets.sh --self-test
-	python3 ./scripts/check_test_wiring.py --self-test
-	python3 ./scripts/check_security_coverage.py --self-test
-	python3 ./scripts/check_scheduled_validation_freshness.py --self-test
-	python3 ./scripts/s3-tests/test_report_compat.py
+	$(RUSTFS_PYTHON_BIN) ./scripts/check_test_wiring.py --self-test
+	$(RUSTFS_PYTHON_BIN) ./scripts/check_security_coverage.py --self-test
+	$(RUSTFS_PYTHON_BIN) ./scripts/check_scheduled_validation_freshness.py --self-test
+	$(RUSTFS_PYTHON_BIN) ./scripts/s3-tests/test_report_compat.py
 	bash -n ./scripts/validate_object_data_cache_cold_stampede.sh
-	python3 ./scripts/check_object_data_cache_follower_samples.py --self-test
+	$(RUSTFS_PYTHON_BIN) ./scripts/check_object_data_cache_follower_samples.py --self-test
 	./scripts/validate_object_data_cache_cold_stampede.sh --self-test

 .PHONY: test
@@ -1,5 +1,7 @@
 # nextest configuration for RustFS.
 #
+experimental = ["setup-scripts"]
+
 # Serialize the ecstore tests that share the process-wide disk registry or
 # exercise a multi-disk commit handoff across nextest process boundaries.
 #
@@ -44,7 +46,36 @@ e2e-reliability = { max-threads = 1 }
 e2e-inline-boundaries = { max-threads = 1 }
 e2e-cluster-nightly = { max-threads = 1 }

+# Deep async storage futures are composed into tests across several crates.
+# Keep the test stack bounded but above libtest's 2 MiB default.
+[scripts.setup.ecstore-base-stack]
+command = ['sh', '-c', 'echo RUST_MIN_STACK=4194304 >> "$NEXTEST_ENV"']
+
+# These exact regression scenarios build deep async storage futures that exceed
+# libtest's 2 MiB spawned-thread stack on Linux. Give only their test processes
+# the same 32 MiB stack already used by the crate's dedicated large-stack tests.
+[scripts.setup.ecstore-large-stack]
+command = ['sh', '-c', 'echo RUST_MIN_STACK=33554432 >> "$NEXTEST_ENV"']
+
+# The serial ILM selection builds the same deep storage futures in both the
+# lifecycle transition module and scanner integration binary. Different tests
+# in each have overflowed first across otherwise unrelated CI runs.
+[scripts.setup.lifecycle-large-stack]
+command = ['sh', '-c', 'echo RUST_MIN_STACK=33554432 >> "$NEXTEST_ENV"']
+
 # --- default profile (local): serialize the flaky groups, never retry --------
+[[profile.default.scripts]]
+filter = 'package(rustfs-ecstore) & test(/^(bucket::lifecycle::bucket_lifecycle_ops::tests::manual_transition_worker_result_recovery_marks_unknown_for_corrupt_marker|services::rebalance::entry::tests::real_rebalance_run_fence_loss_blocks_multipart_publication|store::init::tests::(batch_transitioned_delete_uses_free_version_per_item|decommission_entry_(allows_free_version_consumed_before_source_lock|rejects_subquorum_free_version_conflict_and_retains_source|skips_cleanup_only_marker_when_free_version_is_present)|dispatched_tier_delete_recovery_(checks_later_pool_then_commits_after_source_removal|finds_directory_source_on_encoded_set|retains_journal_on_source_metadata_error)|force_tier_remove_blocks_on_physical_free_version_hidden_by_other_pool|legacy_unknown_transition_delete_falls_back_for_single_batch_and_blocks_prefix|multi_pool_(recursive_prefix_rejects_legacy_or_hidden_merge_loser_before_delete|same_remote_tuple_(batch|single)_delete_waits_for_all_sources|same_tuple_recursive_prefix_uses_one_journal_owner|transitioned_delete_persists_one_free_version_per_remote_tuple)|recursive_prefix_partial_(pool|set)_failure_keeps_prepared_cleanup_owners|restored_transitioned_delete_uses_free_version_as_cleanup_owner|stable_transitioned_recursive_prefix_delete_uses_journal_owners|suspended_null_transition_delete_uses_free_version_as_sole_owner|tier_mutation_peer_handler_applies_prepare_commit_and_abort_idempotently|transition_response_loss_persists_unknown_outcome_for_provider_recovery|transition_transaction_recovery_(drops_record_after_confirmed_local_commit|keeps_cleanup_pending_local_commit)|transitioned_delete_(free_version_replays_after_store_restart|local_quorum_failure_rolls_back_without_cleanup_owner|uses_free_version_as_cleanup_owner)|versioned_delete_marker_keeps_transitioned_source_and_remote_object|versioned_explicit_transition_delete_preserves_other_version_then_allows_bucket_delete))$/)'
+setup = 'ecstore-large-stack'
+
+[[profile.default.scripts]]
+filter = 'package(rustfs-ecstore) | package(rustfs-s3select-api) | package(rustfs-scanner) | (package(rustfs) & test(/^(app::multipart_usecase::tests::concurrent_completions_share_durable_bucket_quota_reservations|app::object::delete::tests::compressed_delete_requests_update_observed_usage_without_releasing_quota_floor|storage::access::tests::(delete_object_access_captures_authorized_bucket_incarnation|copy_operations_reject_recreated_source_bucket_after_authorization|request_slot_keeps_bucket_policy_bound_to_its_store))$/))'
+setup = 'ecstore-base-stack'
+
+[[profile.default.scripts]]
+filter = 'binary(lifecycle_integration_test) | (package(rustfs) & test(/^app::lifecycle_transition_api_test::/))'
+setup = 'lifecycle-large-stack'
+
 [[profile.default.overrides]]
 filter = 'package(rustfs-ecstore) & (test(concurrent_resend_same_part_commits_one_generation) | test(concurrent_config_writes_from_separate_nodes_do_not_lose_writes) | test(/^store::bucket::tests::bucket_delete_(mark_delete|purge_removes|default_s3_delete)/))'
 test-group = 'ecstore-serial-flaky'
@@ -58,6 +89,29 @@ test-group = 'ecstore-serial-flaky'
 filter = 'package(rustfs-ecstore) & test(/^set_disk::ops::multipart::tests::crash_consistency::/)'
 test-group = 'ecstore-serial-flaky'

+# Serialize the heal result-report tests. Every test in the module builds a
+# real-disk (TempDir-backed) hermetic erasure set and drives MiB-scale writes
+# plus deep-scan heal — the same load-sensitive cross-disk IO shape as the
+# crash_consistency scenarios above. Under a heavily parallel run a single
+# disk's IO can fail while write quorum still holds, which flips per-disk
+# readback and aggregate-outcome assertions nondeterministically (different
+# tests each round; all pass standalone). Preventive serialization only, no
+# retries. The matching ci-profile override is after [profile.ci].
+[[profile.default.overrides]]
+filter = 'package(rustfs-ecstore) & test(/^set_disk::ops::heal::heal_result_report_tests::/)'
+test-group = 'ecstore-serial-flaky'
+
+# Serialize the metadata-cache generation-retirement pair. Both carry
+# #[serial(metadata_cache_invalidation_probe)] — a no-op across nextest's
+# process boundary — and assert get_object_metadata_cache generation
+# semantics on a 4-disk hermetic set, the same load-sensitive shape that
+# forced the transition matrix tests into this group. Preventive
+# serialization only, no retries. The matching ci-profile override is after
+# [profile.ci].
+[[profile.default.overrides]]
+filter = 'package(rustfs-ecstore) & test(retires_cached_snapshot)'
+test-group = 'ecstore-serial-flaky'
+
 # The production-handler relocation regression builds an isolated 8-disk,
 # 2-pool store and commits a 72 MiB multipart object. Keep that cross-disk IO
 # from overlapping the ecstore commit fixtures above.
@@ -78,12 +132,29 @@ test-group = 'embedded-test-ports'
 filter = 'package(rustfs-ecstore) & test(manual_transition_page_checkpoint_persists_durable_job_progress)'
 test-group = 'ecstore-serial-flaky'

+# Serialize the transition matrix tests. They build a 4-disk hermetic erasure
+# set, populate the get_object_metadata_cache, and assert generation lifecycle
+# semantics. serial_test's #[serial] has no effect across nextest's process
+# boundary, so concurrent execution races the shared metadata-cache generation
+# counter and causes spurious "metadata read should publish the generation"
+# panics. Preventive serialization, no retries.
+[[profile.default.overrides]]
+filter = 'package(rustfs-ecstore) & test(set_disk::transition_matrix_tests::)'
+test-group = 'ecstore-serial-flaky'
+
 # The durable ILM decommission regressions build isolated multi-pool stores and
 # deliberately take source or target disks offline while checking fencing.
 [[profile.default.overrides]]
 filter = 'package(rustfs-ecstore) & (test(decommission_migrates_and_verifies_registered_durable_ilm_records) | test(decommission_durable_ilm_target_read_error_is_not_masked_by_peer_success) | test(decommission_durable_ilm_terminal_receipt_recovers_failed_source_cleanup) | test(decommission_durable_ilm_receipt_pagination_fails_closed_on_second_page) | test(decommission_durable_ilm_recovery_keeps_multiple_active_sources))'
 test-group = 'ecstore-serial-flaky'

+# Decommission entry and marker/barrier tests share process-wide fault hooks and
+# deterministic commit barriers. Keep the whole init decommission family in one
+# nextest group; serial_test alone cannot isolate separate test processes.
+[[profile.default.overrides]]
+filter = 'package(rustfs-ecstore) & test(/^store::init::tests::(decommission_|suspended_.*decommission)$/)'
+test-group = 'ecstore-serial-flaky'
+
 # Serialize the bucket-incarnation / lifecycle-fence tests. They drive
 # init_bucket_metadata_sys and bucket_metadata_sys_of, i.e. process-global
 # OnceLock state that serial_test's #[serial] cannot protect across nextest's
@@ -113,9 +184,10 @@ filter = 'package(e2e_test) & test(/^inline_fast_path_cluster_test::/)'
 test-group = 'e2e-inline-boundaries'

 # Vault KMS tests share the fixed dev-server port 8200. serial_test's #[serial]
-# does not cross nextest process boundaries, so keep these tests in one group.
+# does not cross nextest process boundaries, so keep every Vault-backed test in
+# one group.
 [[profile.default.overrides]]
-filter = 'package(e2e_test) & test(/^kms::kms_vault_test::/)'
+filter = 'package(e2e_test) & (test(/^kms::kms_vault_test::/) | test(/^kms::kms_rekey_sweep_test::/) | test(/^kms::configured_roundtrip_test::test_configured_vault_kms_admin_and_versioned_cleanup$/))'
 test-group = 'e2e-vault'

 # ---------------------------------------------------------------------------
@@ -133,6 +205,18 @@ fail-fast = false
 # marker is the observable signal the flake policy is built around.
 path = "junit.xml"

+[[profile.ci.scripts]]
+filter = 'package(rustfs-ecstore) & test(/^(bucket::lifecycle::bucket_lifecycle_ops::tests::manual_transition_worker_result_recovery_marks_unknown_for_corrupt_marker|services::rebalance::entry::tests::real_rebalance_run_fence_loss_blocks_multipart_publication|store::init::tests::(batch_transitioned_delete_uses_free_version_per_item|decommission_entry_(allows_free_version_consumed_before_source_lock|rejects_subquorum_free_version_conflict_and_retains_source|skips_cleanup_only_marker_when_free_version_is_present)|dispatched_tier_delete_recovery_(checks_later_pool_then_commits_after_source_removal|finds_directory_source_on_encoded_set|retains_journal_on_source_metadata_error)|force_tier_remove_blocks_on_physical_free_version_hidden_by_other_pool|legacy_unknown_transition_delete_falls_back_for_single_batch_and_blocks_prefix|multi_pool_(recursive_prefix_rejects_legacy_or_hidden_merge_loser_before_delete|same_remote_tuple_(batch|single)_delete_waits_for_all_sources|same_tuple_recursive_prefix_uses_one_journal_owner|transitioned_delete_persists_one_free_version_per_remote_tuple)|recursive_prefix_partial_(pool|set)_failure_keeps_prepared_cleanup_owners|restored_transitioned_delete_uses_free_version_as_cleanup_owner|stable_transitioned_recursive_prefix_delete_uses_journal_owners|suspended_null_transition_delete_uses_free_version_as_sole_owner|tier_mutation_peer_handler_applies_prepare_commit_and_abort_idempotently|transition_response_loss_persists_unknown_outcome_for_provider_recovery|transition_transaction_recovery_(drops_record_after_confirmed_local_commit|keeps_cleanup_pending_local_commit)|transitioned_delete_(free_version_replays_after_store_restart|local_quorum_failure_rolls_back_without_cleanup_owner|uses_free_version_as_cleanup_owner)|versioned_delete_marker_keeps_transitioned_source_and_remote_object|versioned_explicit_transition_delete_preserves_other_version_then_allows_bucket_delete))$/)'
+setup = 'ecstore-large-stack'
+
+[[profile.ci.scripts]]
+filter = 'package(rustfs-ecstore) | package(rustfs-s3select-api) | package(rustfs-scanner) | (package(rustfs) & test(/^(app::multipart_usecase::tests::concurrent_completions_share_durable_bucket_quota_reservations|app::object::delete::tests::compressed_delete_requests_update_observed_usage_without_releasing_quota_floor|storage::access::tests::(delete_object_access_captures_authorized_bucket_incarnation|copy_operations_reject_recreated_source_bucket_after_authorization|request_slot_keeps_bucket_policy_bound_to_its_store))$/))'
+setup = 'ecstore-base-stack'
+
+[[profile.ci.scripts]]
+filter = 'binary(lifecycle_integration_test) | (package(rustfs) & test(/^app::lifecycle_transition_api_test::/))'
+setup = 'lifecycle-large-stack'
+
 # ===========================================================================
 # QUARANTINE — flaky tests granted retries = 2 under the ci profile ONLY.
 #
@@ -165,6 +249,15 @@ test-group = 'ecstore-serial-flaky'
 filter = 'package(rustfs-ecstore) & test(walk_dir_does_not_charge_consumer_backpressure_to_the_stall_budget)'
 retries = 2

+# Serialize the relocated-pool GET resume regression under the ci profile too
+# (see the matching default-profile override near the top). No longer a
+# quarantine: the fixture race (rustfs#6701/rustfs#6703) was fixed by #6707,
+# which made the staging tolerate quorum-tolerated disk gaps; only the 8-disk
+# cross-disk-IO serialization remains.
+[[profile.ci.overrides]]
+filter = 'package(rustfs) & test(execute_get_object_resumes_from_relocated_pool_without_splicing_body)'
+test-group = 'ecstore-serial-flaky'
+
 # Serialize the 4-disk reliability / degraded-read e2e tests under the ci
 # profile too (see the e2e-reliability test-group note near the top). Not a
 # quarantine: no retries, just single-threaded so several 4-disk servers never
@@ -180,8 +273,18 @@ test-group = 'e2e-reliability'
 filter = 'package(rustfs-ecstore) & test(/^set_disk::ops::multipart::tests::crash_consistency::/)'
 test-group = 'ecstore-serial-flaky'

+# Serialize the heal result-report tests under the ci profile too (see the
+# matching default-profile override near the top). Not a quarantine: no
+# retries, just serialized real-disk heal IO.
 [[profile.ci.overrides]]
-filter = 'package(rustfs) & test(execute_get_object_resumes_from_relocated_pool_without_splicing_body)'
+filter = 'package(rustfs-ecstore) & test(/^set_disk::ops::heal::heal_result_report_tests::/)'
+test-group = 'ecstore-serial-flaky'
+
+# Serialize the metadata-cache generation-retirement pair under the ci
+# profile too (see the matching default-profile override near the top). Not a
+# quarantine: no retries.
+[[profile.ci.overrides]]
+filter = 'package(rustfs-ecstore) & test(retires_cached_snapshot)'
 test-group = 'ecstore-serial-flaky'

 # Match the default-profile embedded test isolation without quarantining or
@@ -196,10 +299,20 @@ test-group = 'embedded-test-ports'
 filter = 'package(rustfs-ecstore) & test(manual_transition_page_checkpoint_persists_durable_job_progress)'
 test-group = 'ecstore-serial-flaky'

+# Serialize the transition matrix tests under the ci profile too (see the
+# matching default-profile override near the top). No retries.
+[[profile.ci.overrides]]
+filter = 'package(rustfs-ecstore) & test(set_disk::transition_matrix_tests::)'
+test-group = 'ecstore-serial-flaky'
+
 [[profile.ci.overrides]]
 filter = 'package(rustfs-ecstore) & (test(decommission_migrates_and_verifies_registered_durable_ilm_records) | test(decommission_durable_ilm_target_read_error_is_not_masked_by_peer_success) | test(decommission_durable_ilm_terminal_receipt_recovers_failed_source_cleanup) | test(decommission_durable_ilm_receipt_pagination_fails_closed_on_second_page) | test(decommission_durable_ilm_recovery_keeps_multiple_active_sources))'
 test-group = 'ecstore-serial-flaky'

+[[profile.ci.overrides]]
+filter = 'package(rustfs-ecstore) & test(/^store::init::tests::(decommission_|suspended_.*decommission)$/)'
+test-group = 'ecstore-serial-flaky'
+
 # Serialize the bucket-incarnation / lifecycle-fence tests under the ci profile
 # too (see the matching default-profile override near the top). No retries.
 [[profile.ci.overrides]]
@@ -242,7 +355,8 @@ test-group = 'ecstore-serial-flaky'
 # allowlist", so any new replication test lands in nightly by default (never
 # silently unrun) until it is explicitly blessed as fast here. Keep the two
 # regexes byte-identical. The committed profile selection digests make changes
-# visible in CI; current counts live in docs/testing/e2e-suite-inventory.md.
+# visible in CI; list current membership with `cargo nextest list -p e2e_test
+# --profile <profile>` (platform-dependent; see docs/testing/README.md).
 # HISTORY (2026-07-11): the 20 fast tests were briefly pulled out of this lane
 # (#4724) because they set a loopback (127.0.0.1) replication target that the
 # SSRF egress guard rejected on every PR after repl-1 (#4712). That is fixed —
@@ -364,7 +478,7 @@ path = "junit.xml"
 [profile.e2e-nightly]
 default-filter = """
    package(e2e_test)
-    & test(/^(admin_timeout_regression_test|cluster_concurrency_test|cluster_multidrive_pool_test|heal_erasure_disk_rebuild_test|namespace_lock_quorum_test|object_lambda_test|stale_multipart_cleanup_cluster_test)::/)
+    & test(/^(admin_timeout_regression_test|cluster_concurrency_test|cluster_multidrive_pool_test|degraded_listing_availability_test|heal_erasure_disk_rebuild_test|namespace_lock_quorum_test|object_lambda_test|stale_multipart_cleanup_cluster_test)::/)
 """
 fail-fast = false

@@ -395,7 +509,7 @@ path = "junit.xml"
 # quota, checksum, encryption,
 # security-boundary, ... — that the fast PR `e2e-smoke` subset deliberately
 # skips. Budget <= 45 min; authority for the suite count is `cargo nextest list
-# --profile e2e-full` (see docs/testing/e2e-suite-inventory.md).
+# --profile e2e-full -p e2e_test` (platform-dependent; see docs/testing/README.md).
 #
 # The filter is "the whole e2e_test crate MINUS the sets owned by other lanes":
 #   * protocols:: — FTPS/SFTP/WebDAV, run from the dedicated protocol profile
@@ -409,30 +523,19 @@ path = "junit.xml"
 #     `e2e-smoke` (20 fast) and `e2e-repl-nightly` (56 slow) lanes and reserves
 #     it for those, so e2e-full does not double-run it.
 #   * #[ignore]d tests — nextest skips them by default (no --run-ignored); the
-#     manual-localhost:9000 reliant/policy tests are ci-13's migration.
+#     manual-localhost:9000 reliant tests are ci-13's migration.
 #
 # Each e2e test spawns its own single-node rustfs server on a random port with
 # an isolated temp dir (crates/e2e_test/src/common.rs), so the set is
 # parallel-safe — the same property e2e-smoke relies on. The exceptions are the
 # 4-disk reliability / degraded-read fault-injection tests and the fixed-port
 # Vault tests, both serialized below.
-# KNOWN-FAILURE EXCLUSIONS (characterization run 29381309848, 2026-07-15:
-# 341 ran / 32 failed on the suites' first automated run ever). Deterministic
-# product failures cannot be quarantined away with retries, so each family is
-# excluded here with its tracking issue, under the same discipline as the
-# ci-profile quarantine (docs/testing/README.md): every entry MUST cite one
-# OPEN issue, and the fixing PR MUST delete the exclusion. The passing
-# negative-path siblings of each family stay in as regression guards.
-#   * rustfs#4843 — over-limit archive entry paths hard-reject the whole
-#     archive even under ignore-errors semantics.
 [profile.e2e-full]
 default-filter = """
    package(e2e_test)
    & !test(/^protocols::/)
-    & !test(/^(admin_timeout_regression_test|cluster_concurrency_test|cluster_multidrive_pool_test|heal_erasure_disk_rebuild_test|namespace_lock_quorum_test|object_lambda_test|stale_multipart_cleanup_cluster_test)::/)
+    & !test(/^(admin_timeout_regression_test|cluster_concurrency_test|cluster_multidrive_pool_test|degraded_listing_availability_test|heal_erasure_disk_rebuild_test|namespace_lock_quorum_test|object_lambda_test|stale_multipart_cleanup_cluster_test)::/)
    & !test(/^replication_extension_test::/)
-    & !test(/^multipart_auth_test::test_signed_put_object_extract_skips_invalid_entry_when_ignore_errors_enabled$/)
-    & !test(/^snowball_auto_extract_test::tests::snowball_auto_extract_(ignores_invalid_entries_when_requested|supports_standard_headers_with_combined_extract_options)$/)
 """
 fail-fast = false

@@ -453,5 +556,5 @@ filter = 'package(e2e_test) & test(/^inline_fast_path_cluster_test::/)'
 test-group = 'e2e-inline-boundaries'

 [[profile.e2e-full.overrides]]
-filter = 'package(e2e_test) & test(/^kms::kms_vault_test::/)'
+filter = 'package(e2e_test) & (test(/^kms::kms_vault_test::/) | test(/^kms::kms_rekey_sweep_test::/) | test(/^kms::configured_roundtrip_test::test_configured_vault_kms_admin_and_versioned_cleanup$/))'
 test-group = 'e2e-vault'
@@ -9,4 +9,4 @@
 # if the selected count drops below this number, so a rename or removal that
 # thins the security smoke gate must update this file in the same PR.
 # Adding tests does not require a bump, but bumping keeps the guard tight.
-16
+18
@@ -0,0 +1,84 @@
+# Copyright 2024 RustFS Team
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#     http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+
+global:
+  scrape_interval: 15s # Evaluate rules every 15 seconds. The default is every 1 minute.
+  evaluation_interval: 15s
+  external_labels:
+    cluster: 'rustfs-dev'  # Label to identify the cluster
+    replica: '1'  # Replica identifier
+
+rule_files:
+  - /etc/prometheus/rules/*.yml
+
+scrape_configs:
+  - job_name: 'otel-collector'
+    static_configs:
+      - targets: [ 'otel-collector:8888' ]  # Scrape metrics from Collector
+    scrape_interval: 10s
+
+  - job_name: 'rustfs-app-metrics'
+    static_configs:
+      - targets: [ 'otel-collector:8889' ]  # Application indicators
+    scrape_interval: 15s
+    metric_relabel_configs:
+      - source_labels: [ __name__ ]
+        regex: 'go_.*'
+        action: drop  # Drop Go runtime metrics if not needed
+
+  - job_name: 'tempo'
+    static_configs:
+      - targets: [ 'tempo:3200' ]  # Scrape metrics from Tempo
+
+  - job_name: 'jaeger'
+    static_configs:
+      - targets: [ 'jaeger:14269' ]  # Jaeger admin port (14269 is standard for admin/metrics)
+
+  - job_name: 'loki'
+    static_configs:
+      - targets: [ 'loki:3100' ]
+
+  - job_name: 'prometheus'
+    static_configs:
+      - targets: [ 'localhost:9090' ]
+
+  - job_name: 'vulture'
+    static_configs:
+      - targets:
+          - 'vulture:8080'
+
+otlp:
+  promote_resource_attributes:
+    - service.instance.id
+    - service.name
+    - service.namespace
+    - cloud.availability_zone
+    - cloud.region
+    - container.name
+    - deployment.environment.name
+    - k8s.cluster.name
+    - k8s.container.name
+    - k8s.cronjob.name
+    - k8s.daemonset.name
+    - k8s.deployment.name
+    - k8s.job.name
+    - k8s.namespace.name
+    - k8s.pod.name
+    - k8s.replicaset.name
+    - k8s.statefulset.name
+  translation_strategy: NoUTF8EscapingWithSuffixes
+
+storage:
+  tsdb:
+    out_of_order_time_window: 30m
@@ -5,3 +5,5 @@ self-hosted-runner:
    - sm-standard-2
    - sm-standard-4
    - dind-sm-standard-2
+    - smoke-testing
+    - pf-testing
@@ -7,8 +7,16 @@
  { "workflow": ".github/workflows/e2e-s3tests.yml", "max_age_hours": 192 },
  { "workflow": ".github/workflows/fuzz.yml", "max_age_hours": 36 },
  { "workflow": ".github/workflows/mint.yml", "max_age_hours": 192 },
-  { "workflow": ".github/workflows/minio-interop.yml", "max_age_hours": 36 },
+  {
+    "workflow": ".github/workflows/minio-interop.yml",
+    "max_age_hours": 36,
+    "never_ran_grace_until": "2026-09-08T00:00:00Z"
+  },
  { "workflow": ".github/workflows/nightly-gnu.yml", "max_age_hours": 36 },
  { "workflow": ".github/workflows/performance-ab.yml", "max_age_hours": 36 },
-  { "workflow": ".github/workflows/runner-hygiene.yml", "max_age_hours": 792 }
+  {
+    "workflow": ".github/workflows/runner-hygiene.yml",
+    "max_age_hours": 792,
+    "never_ran_grace_until": "2026-09-02T06:37:00Z"
+  }
 ]
@@ -24,8 +24,11 @@ on:
      - '.github/actions/**'
      - '.github/workflows/**'
      - 'scripts/release/create_or_update_release.sh'
+      - 'scripts/release/package_versions.sh'
+      - 'scripts/test_package_versions.sh'
      - 'scripts/security/check_performance_ab_workflow.sh'
      - 'scripts/security/check_preview_release_workflow.sh'
+      - 'scripts/security/check_tier_artifact_workflow.sh'
      - 'scripts/security/check_workflow_pins.sh'
  pull_request:
    types: [ opened, synchronize, reopened, closed ]
@@ -37,8 +40,11 @@ on:
      - '.github/actions/**'
      - '.github/workflows/**'
      - 'scripts/release/create_or_update_release.sh'
+      - 'scripts/release/package_versions.sh'
+      - 'scripts/test_package_versions.sh'
      - 'scripts/security/check_performance_ab_workflow.sh'
      - 'scripts/security/check_preview_release_workflow.sh'
+      - 'scripts/security/check_tier_artifact_workflow.sh'
      - 'scripts/security/check_workflow_pins.sh'
  schedule:
    # Daily, not weekly. This schedule exists to catch RustSec advisories
@@ -146,6 +152,12 @@ jobs:
      - name: Check performance A/B workflow trust boundary
        run: ./scripts/security/check_performance_ab_workflow.sh

+      - name: Check tier evidence workflow isolation
+        run: ./scripts/security/check_tier_artifact_workflow.sh
+
+      - name: Check package version contract
+        run: ./scripts/test_package_versions.sh
+
  dependency-review:
    name: Dependency Review
    runs-on: ubuntu-latest
@@ -244,7 +244,7 @@ jobs:
    needs: [ build-check, prepare-platform-matrix ]
    if: needs.build-check.outputs.should_build == 'true' && needs.prepare-platform-matrix.result == 'success'
    runs-on: ${{ matrix.os }}
-    timeout-minutes: 150
+    timeout-minutes: 180
    env:
      FORCE_JAVASCRIPT_ACTIONS_TO_NODE24: "true"
      # Release binaries ship without dial9 telemetry and therefore do not need
@@ -408,9 +408,9 @@ jobs:

          if [[ "${{ matrix.cross }}" == "true" ]]; then
            # All cross targets in the matrix are Linux; zigbuild handles them.
-            cargo zigbuild --release --target ${{ matrix.target }} -p rustfs --bins
+            cargo zigbuild --release --target ${{ matrix.target }} -p rustfs --bin rustfs
          else
-            cargo build --release --target ${{ matrix.target }} -p rustfs --bins
+            cargo build --release --target ${{ matrix.target }} -p rustfs --bin rustfs
          fi

      - name: Create release package
@@ -1033,6 +1033,55 @@ jobs:
          echo "🎉 Released $TAG successfully!"
          echo "📄 Release URL: ${{ needs.create-release.outputs.release_url }}"

+  # Remove the internal preview releases once the deliverable release is live.
+  # Only the Releases are deleted; the -preview.N tags stay so the validated
+  # commit remains traceable.
+  cleanup-preview-releases:
+    name: Cleanup Preview Releases
+    needs: [ build-check, publish-release ]
+    if: startsWith(github.ref, 'refs/tags/') && (needs.build-check.outputs.build_type == 'release' || needs.build-check.outputs.build_type == 'prerelease')
+    runs-on: ubuntu-latest
+    timeout-minutes: 30
+    permissions:
+      contents: write
+    steps:
+      - name: Delete preview releases for this target
+        env:
+          GH_TOKEN: ${{ github.token }}
+        shell: bash
+        run: |
+          set -euo pipefail
+
+          TAG="${{ needs.build-check.outputs.version }}"
+          RELEASES_JSON="${RUNNER_TEMP}/releases.json"
+
+          # Fetch before filtering: a failed listing must abort here instead of
+          # looking like "nothing to clean up".
+          gh api --paginate "repos/${GITHUB_REPOSITORY}/releases?per_page=100" > "$RELEASES_JSON"
+
+          # Match only <target>-preview.<digits>. String operations, not a
+          # regex over the tag, so dots in the version cannot widen the match.
+          DELETED=0
+          while IFS= read -r preview_tag; do
+            [[ -n "$preview_tag" ]] || continue
+            echo "🧹 Deleting preview release $preview_tag (tag kept)"
+            gh release delete "$preview_tag" --repo "${GITHUB_REPOSITORY}" --yes
+            DELETED=$((DELETED + 1))
+          done < <(
+            jq -r --arg tag "$TAG" '
+              .[]
+              | select(.tag_name | startswith($tag + "-preview."))
+              | select(.tag_name | ltrimstr($tag + "-preview.") | test("^[0-9]+$"))
+              | .tag_name
+            ' "$RELEASES_JSON"
+          )
+
+          if [[ "$DELETED" -eq 0 ]]; then
+            echo "ℹ️  No preview releases to clean up for $TAG"
+          else
+            echo "✅ Removed $DELETED preview release(s) for $TAG"
+          fi
+
  alert-on-failure:
    name: Alert on scheduled failure
    needs: [build-check, prepare-platform-matrix, build-rustfs, build-summary]
@@ -212,6 +212,9 @@ jobs:
          install-build-packaging-tools: 'false'

      - name: Build ci-feat-rio superset
+        env:
+          # --all-targets links the same test binaries as the reader lane.
+          CARGO_BUILD_JOBS: "2"
        run: |
          cargo build -p rustfs -p rustfs-ecstore --all-targets --features rio-v2
          cargo build -p rustfs --bins --features rio-v2,e2e-test-hooks
@@ -240,6 +243,9 @@ jobs:
          install-build-packaging-tools: 'false'

      - name: Build ci-feat-proto superset
+        env:
+          # Avoid an unbounded burst of concurrent test-binary links (#5394).
+          CARGO_BUILD_JOBS: "2"
        run: |
          cargo build -p rustfs -p rustfs-protocols --all-targets --features swift
          cargo build -p rustfs -p rustfs-protocols --all-targets --features sftp
@@ -142,6 +142,9 @@ jobs:
      - name: Check logging guardrails
        run: ./scripts/check_logging_guardrails.sh

+      - name: Check error other(format!) ratchet
+        run: ./scripts/check_error_other_format_ratchet.sh
+
      - name: Check tokio io-uring feature guard
        run: ./scripts/check_no_tokio_io_uring.sh

@@ -181,22 +184,14 @@ jobs:
    needs: [ quick-checks ]
    runs-on: sm-standard-4
    timeout-minutes: 90
-    # Both lines are required. Job-level `permissions` replaces the workflow
-    # block rather than merging with it, so declaring only `actions: write`
-    # would drop `contents: read` and break this job's checkout and the
-    # repo-token the setup action hands to setup-protoc.
-    permissions:
-      contents: read
-      actions: write
    env:
      FORCE_JAVASCRIPT_ACTIONS_TO_NODE24: "true"
    steps:
      - name: Checkout repository
        uses: actions/checkout@9c091bb21b7c1c1d1991bb908d89e4e9dddfe3e0 # v7
        with:
-          # This job's token can cancel runs and delete Actions caches. Checkout
-          # otherwise writes it into .git/config, where a PR's own build.rs or
-          # proc-macro could read it back out.
+          # Checkout otherwise writes the token into .git/config, where a PR's
+          # own build.rs or proc-macro could read it back out.
          persist-credentials: false

      - name: Setup Rust environment
@@ -212,6 +207,9 @@ jobs:
          cache-save-if: 'false'
          install-build-packaging-tools: 'false'

+      - name: Protect Connect test home
+        run: chmod go-w "$(realpath "$HOME")"
+
      - name: Prepare test evidence
        run: |
          mkdir -p artifacts/test-and-lint
@@ -310,6 +308,9 @@ jobs:
          } > artifacts/test-and-lint/doctest-diagnostics.txt
          exit "${status}"

+      - name: Check offline enrollment E2E root boundary
+        run: ./scripts/check_offline_enrollment_e2e.sh
+
      - name: Upload test reports and diagnostics
        if: always()
        uses: actions/upload-artifact@b7c566a772e6b6bfb58ed0dc250532a479d7789f # v6
@@ -344,41 +345,36 @@ jobs:
      - name: Run rebalance/decommission migration proofs
        run: ./scripts/check_migration_gate_count.sh

-      # Early stop. Once this job has failed the PR cannot merge, so the sibling
-      # lanes are burning runners on a result nobody can act on: on run
-      # 30674613104 three lanes had already failed while Test and Lint and the
-      # rio-v2 variant kept going past 70 minutes.
-      #
-      # Only this job may cancel. The lanes that are NOT required checks
-      # (protocols, ILM, e2e, s3-tests) must never hold that power: a flake in
-      # one of them would turn the required "Test and Lint" into `cancelled`,
-      # which blocks the merge. Today a maintainer can merge with sftp red, and
-      # that has to stay true.
-      #
-      # These steps run last so the `if: always()` artifact upload above still
-      # captures logs and diagnostics before the run goes away.
+      # Record the reason before this job completes as FAILURE. A separate
+      # dependent job cancels sibling lanes only after GitHub has preserved this
+      # required check's failure verdict.
      - name: Annotate early-stop reason
-        if: failure() && github.event_name == 'pull_request'
-        run: |
-          {
-            echo "## CI early-stop"
-            echo "Job \`${GITHUB_JOB}\` (Test and Lint) failed; cancelling run ${GITHUB_RUN_ID} to free runners."
-            echo "Sibling jobs showing **cancelled** were stopped by this job, not by their own failure."
-          } >> "$GITHUB_STEP_SUMMARY"
-
-      # curl rather than `gh`: every existing `gh` call in this repo runs on
-      # ubuntu-latest, and the sm-standard-* images are custom and trimmed (they
-      # ship no C toolchain, see the e2e job below), so `gh` is not known to
-      # exist here.
-      #
-      # Fork PRs are excluded explicitly instead of relying on the error path:
-      # their GITHUB_TOKEN is forced read-only and job-level permissions cannot
-      # raise it, so the call would always 403. Skipping keeps their logs clean.
-      - name: Cancel run on failure (same-repo PR only)
        if: >-
          failure() && github.event_name == 'pull_request'
          && github.event.pull_request.head.repo.full_name == github.repository
-        continue-on-error: true
+        run: |
+          {
+            echo "## CI early-stop"
+            echo "Job \`${GITHUB_JOB}\` (Test and Lint) failed; a follow-up job will cancel sibling lanes to free runners."
+            echo "Sibling jobs showing **cancelled** were stopped by the early-stop follow-up, not by their own failure."
+          } >> "$GITHUB_STEP_SUMMARY"
+
+  # Preserve the required Test and Lint FAILURE verdict before stopping sibling
+  # lanes. Cancelling from inside test-and-lint changed its own conclusion to
+  # CANCELLED and hid the actionable failure in the PR checks UI.
+  cancel-after-test-and-lint-failure:
+    name: Cancel siblings after Test and Lint failure
+    if: >-
+      failure() && needs.test-and-lint.result == 'failure'
+      && github.event_name == 'pull_request'
+      && github.event.pull_request.head.repo.full_name == github.repository
+    needs: [ test-and-lint ]
+    runs-on: ubuntu-latest
+    timeout-minutes: 5
+    permissions:
+      actions: write
+    steps:
+      - name: Cancel remaining jobs
        env:
          GH_TOKEN: ${{ secrets.GITHUB_TOKEN }}
        run: |
@@ -386,7 +382,7 @@ jobs:
            -H "Authorization: Bearer ${GH_TOKEN}" \
            -H "Accept: application/vnd.github+json" \
            -H "X-GitHub-Api-Version: 2022-11-28" \
-            "${GITHUB_API_URL}/repos/${GITHUB_REPOSITORY}/actions/runs/${GITHUB_RUN_ID}/cancel" || true
+            "${GITHUB_API_URL}/repos/${GITHUB_REPOSITORY}/actions/runs/${GITHUB_RUN_ID}/cancel"

  # Dedicated serial lane for the ILM / lifecycle integration tests. These tests
  # drive the object layer through process-global singletons (the GLOBAL_ENV
@@ -433,10 +429,38 @@ jobs:
      #   - test_noncurrent_{expiry,transition}_still_works_after_immediate_compensation_transition:
      #     noncurrent transition/expiry after an immediate compensation transition.
      - name: Run ignored ILM integration tests serially
+        env:
+          # Match the measured Test and Lint link budget. The default exposed
+          # all 14 pod CPUs and a cold cache spent the full 80m compiling
+          # without starting one ILM test (main run 32982910990).
+          CARGO_BUILD_JOBS: ${{ (github.event_name == 'push' || github.event_name == 'workflow_dispatch') && '3' || '2' }}
        run: |
-          cargo nextest run -j1 --run-ignored ignored-only \
+          mkdir -p artifacts/ilm-integration
+          set +e
+          NEXTEST_HIDE_PROGRESS_BAR=1 timeout --verbose --signal=TERM --kill-after=30s 80m \
+            cargo nextest run -j1 --run-ignored ignored-only \
            -p rustfs-scanner -p rustfs \
-            -E '(binary(lifecycle_integration_test) or (package(rustfs) and test(lifecycle_transition_api_test))) and not (test(test_noncurrent_expiry_still_works_after_immediate_compensation_transition) or test(test_noncurrent_transition_still_works_after_immediate_compensation_transition))'
+            -E '(binary(lifecycle_integration_test) or (package(rustfs) and test(lifecycle_transition_api_test))) and not (test(test_noncurrent_expiry_still_works_after_immediate_compensation_transition) or test(test_noncurrent_transition_still_works_after_immediate_compensation_transition))' \
+            --status-level all --final-status-level all \
+            2>&1 | tee artifacts/ilm-integration/nextest.log
+          status=${PIPESTATUS[0]}
+          {
+            echo "exit_status=${status}"
+            echo "finished_at=$(date --utc --iso-8601=seconds)"
+            echo
+            echo "Remaining test-related processes:"
+            pgrep -af 'cargo|nextest|target/.*/deps/' || true
+          } > artifacts/ilm-integration/diagnostics.txt
+          exit "${status}"
+
+      - name: Upload ILM test diagnostics
+        if: always()
+        uses: actions/upload-artifact@b7c566a772e6b6bfb58ed0dc250532a479d7789f # v6
+        with:
+          name: ilm-integration-${{ github.run_number }}-${{ github.run_attempt }}
+          path: |
+            artifacts/ilm-integration
+            target/nextest/ci/junit.xml

  test-and-lint-rio-v2:
    name: Test and Lint (rio-v2)
@@ -460,14 +484,83 @@ jobs:
          cache-save-if: 'false'
          install-build-packaging-tools: 'false'

+      - name: Protect Connect test home
+        run: chmod go-w "$(realpath "$HOME")"
+
      - name: Run rio-v2 clippy lints
        run: cargo clippy -p rustfs -p rustfs-ecstore --all-targets --features rio-v2 -- -D warnings

      - name: Run rio-v2 feature tests
+        env:
+          # Match the main nextest lane's #5394 link-I/O guard. A cold feature
+          # cache otherwise fans out enough rust-lld processes to exhaust this
+          # job's 90-minute budget before any test starts.
+          CARGO_BUILD_JOBS: "2"
        run: |
-          cargo nextest run -p rustfs -p rustfs-ecstore --features rio-v2
+          # --profile ci so the quarantine list (and its junit flaky markers)
+          # covers this leg too; the default profile is the local no-retry
+          # profile and silently ignored quarantined flakes here (rustfs#6703).
+          cargo nextest run --profile ci -p rustfs -p rustfs-ecstore --features rio-v2
          cargo test -p rustfs --doc --features rio-v2

+  connect-short-credential-boundary:
+    name: Connect Short Credential Boundary
+    if: github.event_name != 'pull_request' || github.event.action != 'closed'
+    needs: [ quick-checks ]
+    runs-on: sm-standard-4
+    timeout-minutes: 60
+    env:
+      FORCE_JAVASCRIPT_ACTIONS_TO_NODE24: "true"
+    steps:
+      - name: Checkout repository
+        uses: actions/checkout@9c091bb21b7c1c1d1991bb908d89e4e9dddfe3e0 # v7
+        with:
+          persist-credentials: false
+
+      - name: Setup Rust environment
+        uses: ./.github/actions/setup
+        with:
+          rust-version: stable
+          cache-shared-key: ci-dev
+          cache-save-if: 'false'
+          install-build-packaging-tools: 'false'
+          install-test-tools: 'false'
+
+      - name: Run short credential behavior tests
+        env:
+          CARGO_BUILD_JOBS: "2"
+        run: |
+          cargo test -p rustfs --test connect_registration \
+            --features connect-e2e-short-credentials \
+            registration_enforces_build_profile_credential_lifetime -- --exact
+          cargo test -p rustfs --test connect_registration \
+            --features connect-e2e-short-credentials \
+            rotation_waits_for_threshold_and_stops_on_revocation -- --exact
+
+      - name: Reject short credentials in release builds
+        env:
+          CARGO_BUILD_JOBS: "2"
+        run: |
+          log="$(mktemp)"
+          set +e
+          CARGO_TERM_COLOR=never cargo check -p rustfs --release \
+            --features connect-e2e-short-credentials >"$log" 2>&1
+          status=$?
+          set -e
+          cat "$log"
+          expected='error: connect-e2e-short-credentials is restricted to debug builds'
+          summary="error: could not compile \`rustfs\` (lib) due to 1 previous error"
+          expected_count="$(grep -Fxc "$expected" "$log" || true)"
+          summary_count="$(grep -Fc "$summary" "$log" || true)"
+          error_count="$(grep -Ec '^error(:|\[)' "$log" || true)"
+          if [ "$status" -ne 101 ] || [ "$expected_count" -ne 1 ] \
+            || [ "$summary_count" -ne 1 ] || [ "$error_count" -ne 2 ]; then
+            echo "release feature gate did not fail solely at the expected compile_error" >&2
+            rm -f "$log"
+            exit 1
+          fi
+          rm -f "$log"
+
  test-and-lint-protocols:
    name: "Test and Lint (${{ matrix.features.name }})"
    if: github.event_name != 'pull_request' || github.event.action != 'closed'
@@ -504,13 +597,23 @@ jobs:
          cache-save-if: 'false'
          install-build-packaging-tools: 'false'

+      - name: Protect Connect test home
+        run: chmod go-w "$(realpath "$HOME")"
+
      - name: Run clippy with ${{ matrix.features.name }}
        run: |
          cargo clippy -p rustfs -p rustfs-protocols --all-targets ${{ matrix.features.flags }} -- -D warnings

      - name: Run tests with ${{ matrix.features.name }}
+        env:
+          # Keep feature-test linking under the same bounded concurrency as the
+          # main nextest lane; Clippy is metadata-only and needs no such limit.
+          CARGO_BUILD_JOBS: "2"
        run: |
-          cargo nextest run -p rustfs -p rustfs-protocols ${{ matrix.features.flags }}
+          # --profile ci so the quarantine list (and its junit flaky markers)
+          # covers this leg too; the default profile is the local no-retry
+          # profile and silently ignored quarantined flakes here (rustfs#6703).
+          cargo nextest run --profile ci -p rustfs -p rustfs-protocols ${{ matrix.features.flags }}

  build-rustfs-debug-binary:
    name: Build RustFS Debug Binary
@@ -311,7 +311,7 @@ jobs:
      - name: Wait for RustFS ready
        run: |
          for _ in {1..120}; do
-            if curl -sf "http://${S3_HOST}:${S3_PORT}/health" >/dev/null 2>&1; then
+            if curl -sf "http://${S3_HOST}:${S3_PORT}/health/ready" >/dev/null 2>&1; then
              echo "RustFS is ready"
              exit 0
            fi
@@ -0,0 +1,117 @@
+# Copyright 2024 RustFS Team
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#     http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+
+name: Upgrade Compatibility
+
+on:
+  pull_request:
+    paths:
+      - ".github/workflows/e2e-upgrade.yml"
+      - "crates/e2e_test/src/common.rs"
+      - "crates/e2e_test/src/lib.rs"
+      - "crates/e2e_test/src/upgrade_compatibility_test.rs"
+      - "crates/ecstore/**"
+      - "crates/filemeta/**"
+      - "crates/kms/**"
+      - "crates/storage-api/**"
+      - "rustfs/**"
+      - "Cargo.lock"
+  push:
+    tags:
+      - "[0-9]*.[0-9]*.[0-9]*"
+  schedule:
+    - cron: "17 3 * * 1"
+  workflow_dispatch:
+
+permissions:
+  contents: read
+
+concurrency:
+  group: ${{ github.workflow }}-${{ github.ref }}
+  cancel-in-progress: ${{ github.event_name == 'pull_request' }}
+
+env:
+  CARGO_TERM_COLOR: always
+  RUST_BACKTRACE: 1
+  UPGRADE_SOURCE_VERSION: 1.0.0-rc.2
+  UPGRADE_SOURCE_ASSET: rustfs-linux-x86_64-gnu-v1.0.0-rc.2.zip
+  UPGRADE_SOURCE_SHA256: 7c789386bf85278f865b8e0d359bf4edb84d5aa408cc3fa54a18c25ca74cd6e7
+
+jobs:
+  upgrade:
+    name: ${{ matrix.name }}
+    strategy:
+      fail-fast: false
+      matrix:
+        include:
+          - name: Direct upgrade from rc.2
+            cache_key: e2e-direct-upgrade
+            test: direct_upgrade_from_rc2_preserves_object_contracts
+            artifact: direct-upgrade
+          - name: Mixed-version rolling upgrade from rc.2
+            cache_key: e2e-mixed-version-upgrade
+            test: rolling_upgrade_from_rc2_preserves_mixed_version_contracts
+            artifact: mixed-version-upgrade
+    runs-on: ubuntu-latest
+    timeout-minutes: 60
+    env:
+      FORCE_JAVASCRIPT_ACTIONS_TO_NODE24: "true"
+    steps:
+      - name: Checkout repository
+        uses: actions/checkout@9c091bb21b7c1c1d1991bb908d89e4e9dddfe3e0 # v7
+        with:
+          persist-credentials: false
+
+      - name: Setup Rust environment
+        uses: ./.github/actions/setup
+        with:
+          cache-shared-key: ${{ matrix.cache_key }}
+          cache-save-if: ${{ github.ref == 'refs/heads/main' }}
+          install-build-packaging-tools: "false"
+
+      - name: Download pinned previous release
+        env:
+          SOURCE_DIR: ${{ runner.temp }}/rustfs-upgrade-source
+        run: |
+          set -euo pipefail
+          mkdir -p "$SOURCE_DIR"
+          archive="$SOURCE_DIR/$UPGRADE_SOURCE_ASSET"
+          curl --fail --location --retry 3 --output "$archive" \
+            "https://github.com/${GITHUB_REPOSITORY}/releases/download/${UPGRADE_SOURCE_VERSION}/${UPGRADE_SOURCE_ASSET}"
+          echo "$UPGRADE_SOURCE_SHA256  $archive" | sha256sum --check --strict
+          unzip -q "$archive" -d "$SOURCE_DIR"
+          chmod +x "$SOURCE_DIR/rustfs"
+          test -x "$SOURCE_DIR/rustfs"
+          echo "RUSTFS_UPGRADE_SOURCE_BINARY=$SOURCE_DIR/rustfs" >> "$GITHUB_ENV"
+          echo "RUSTFS_E2E_LOG_DIR=$RUNNER_TEMP/rustfs-upgrade-logs" >> "$GITHUB_ENV"
+
+      - name: Build current RustFS binary
+        run: |
+          cargo build --locked -p rustfs --bin rustfs
+          : > target/debug/rustfs.features
+
+      - name: Run upgrade compatibility test
+        run: |
+          cargo test --locked -p e2e_test \
+            "upgrade_compatibility_test::${{ matrix.test }}" \
+            -- --ignored --exact --nocapture
+
+      - name: Upload server logs
+        if: always()
+        uses: actions/upload-artifact@b7c566a772e6b6bfb58ed0dc250532a479d7789f # v6
+        with:
+          name: ${{ matrix.artifact }}-server-logs-${{ github.run_number }}
+          path: ${{ runner.temp }}/rustfs-upgrade-logs
+          if-no-files-found: warn
+          retention-days: 14
@@ -173,7 +173,7 @@ jobs:
          path: |
            fuzz/artifacts/**
            fuzz/corpus/${{ matrix.target }}/**
-          if-no-files-found: ignore
+          if-no-files-found: error
          retention-days: 7

  # ──────────────────────────────────────────────────────────────
@@ -227,7 +227,7 @@ jobs:
          path: |
            fuzz/artifacts/**
            fuzz/corpus/${{ matrix.target }}/**
-          if-no-files-found: ignore
+          if-no-files-found: error
          retention-days: 30

  # ──────────────────────────────────────────────────────────────
@@ -20,22 +20,27 @@
 # each run with Docker and then runs the `#[ignore]` reader tests in
 # rustfs/src/storage/minio_generated_read_test.rs.
 #
-# Scope: end-to-end MinIO-to-RustFS SSE interop is NOT implemented yet. Both
-# envelope parsers reject MinIO's own wrapped-DEK shape — see
-# `is_data_key_envelope` in crates/kms/src/encryption/dek.rs and the
-# `deny_unknown_fields` `LocalSseDekEnvelope` in rustfs/src/storage/sse.rs — and
-# closing that gap is tracked in rustfs/backlog#1638. Treat this job as the
-# harness for #1638, not as standing evidence that a MinIO migration reads back.
+# Scope: MinIO-to-RustFS SSE read interop is implemented behind the `rio-v2`
+# feature for MinIO's builtin static-KMS deployments — SSE-S3 and SSE-KMS
+# (single- and multipart) since rustfs/rustfs#6191, SSE-C detection since the
+# rustfs/backlog#1638 D2 close-out. This job is the standing evidence: it
+# regenerates real MinIO backend trees and proves byte-identical plaintext
+# reconstruction. KES/MinKMS-backed MinIO objects remain unreadable by design
+# (their envelopes are sealed by the KES service, not by a key RustFS can
+# hold), and default RustFS builds do not include the read path — it is a
+# special-purpose migration capability, not a default-build feature.
 #
 # Runner: GitHub-hosted `ubuntu-latest`. It reliably ships Docker + Python,
 # unlike the self-hosted fleet, whose pods drift in Docker/pip availability
 # (see the infra note in e2e-s3tests.yml). Nightly + manual only.
-# DISABLED. This workflow is switched off in the repository's Actions settings
-# (state: disabled_manually) and does not run on any trigger, including its cron
-# and workflow_dispatch. That state lives in GitHub's UI and is invisible when
-# reading this file, which has already misled at least one audit — hence this
-# banner. Re-enabling is a UI action; anyone doing so should first check that the
-# workflow still matches the current CI layout. See rustfs/backlog#1603.
+#
+# Enablement: this workflow was long disabled in the repository's Actions
+# settings (state: disabled_manually — a state that lives in GitHub's UI and is
+# invisible in this file). The change that updated this banner also re-added
+# the .github/scheduled-validations.json entry; both only make sense together
+# with re-enabling the workflow in the Actions settings. If it is ever disabled
+# again, remove the scheduled-validations entry in the same change — a disabled
+# workflow can never satisfy the freshness check. See rustfs/backlog#1603.
 #
 name: minio-interop

@@ -155,7 +155,7 @@ jobs:
      - name: Wait for RustFS ready
        run: |
          for _ in {1..60}; do
-            if curl -sf http://127.0.0.1:9000/health >/dev/null 2>&1; then
+            if curl -sf http://127.0.0.1:9000/health/ready >/dev/null 2>&1; then
              echo "RustFS is ready"
              exit 0
            fi
@@ -34,7 +34,7 @@ env:
 jobs:
  build:
    name: Build x86_64 GNU
-    runs-on: sm-standard-2
+    runs-on: sm-standard-4
    timeout-minutes: 150
    env:
      FORCE_JAVASCRIPT_ACTIONS_TO_NODE24: "true"
@@ -55,6 +55,155 @@ jobs:
      - name: Build RustFS
        run: cargo build --release --locked --target x86_64-unknown-linux-gnu -p rustfs --bins

+      - name: Build DEB package
+        id: deb
+        shell: bash
+        run: |
+          set -euo pipefail
+
+          # Nightly snapshot name: rustfs-nightly-<YYYY-MM-DD> (Asia/Shanghai,
+          # matching the schedule timezone so the file date always matches the
+          # cron's intended day).
+          DEB_DATE="$(TZ=Asia/Shanghai date +%Y-%m-%d)"
+          DEB_FILE="rustfs-nightly-${DEB_DATE}.deb"
+          PKG_DIR="rustfs-nightly-${DEB_DATE}"
+
+          command -v fakeroot >/dev/null 2>&1 || sudo apt-get install -y -qq fakeroot
+
+          BIN="target/x86_64-unknown-linux-gnu/release/rustfs"
+          test -x "${BIN}" || { echo "rustfs binary not found: ${BIN}"; exit 1; }
+
+          mkdir -p "${PKG_DIR}/DEBIAN"
+          mkdir -p "${PKG_DIR}/usr/bin"
+          mkdir -p "${PKG_DIR}/etc/default"
+          mkdir -p "${PKG_DIR}/lib/systemd/system"
+          mkdir -p "${PKG_DIR}/usr/share/doc/rustfs"
+
+          cp "${BIN}" "${PKG_DIR}/usr/bin/rustfs"
+          chmod 755 "${PKG_DIR}/usr/bin/rustfs"
+          cp deploy/build/rustfs.service "${PKG_DIR}/lib/systemd/system/"
+
+          cat > "${PKG_DIR}/etc/default/rustfs" << 'ENVEOF'
+          # RustFS Environment Configuration
+          # See https://rustfs.com/docs/ for more information
+          # RUSTFS_VOLUMES=""
+          # RUSTFS_ROOT_USER=""
+          # RUSTFS_ROOT_PASSWORD=""
+          ENVEOF
+
+          # dpkg versions must start with a digit and cannot contain hyphens;
+          # a date-based snapshot version keeps the nightly installable
+          # alongside release packages.
+          DEB_VERSION="${DEB_DATE//-/.}~nightly"
+
+          cat > "${PKG_DIR}/DEBIAN/control" << EOF
+          Package: rustfs
+          Version: ${DEB_VERSION}
+          Section: utils
+          Priority: optional
+          Architecture: amd64
+          Depends: libc6 (>= 2.31)
+          Maintainer: RustFS Team <[email protected]>
+          Description: High-performance distributed object storage
+           RustFS is a high-performance distributed object storage software
+           built using Rust. It is compatible with MinIO and S3 API.
+          Homepage: https://rustfs.com
+          EOF
+
+          cat > "${PKG_DIR}/DEBIAN/conffiles" << 'CONFFILES'
+          /etc/default/rustfs
+          CONFFILES
+
+          cat > "${PKG_DIR}/DEBIAN/postinst" << 'POSTINST'
+          #!/bin/bash
+          set -e
+          if ! getent passwd rustfs > /dev/null 2>&1; then
+              useradd -r -s /bin/false -d /opt/rustfs rustfs
+          fi
+          mkdir -p /opt/rustfs /data/rustfs /var/log/rustfs
+          chown rustfs:rustfs /opt/rustfs /data/rustfs /var/log/rustfs
+          if [ -d /run/systemd/system ]; then
+              systemctl daemon-reload
+          fi
+          echo "RustFS installed. Configure /etc/default/rustfs then: systemctl start rustfs"
+          POSTINST
+          chmod 755 "${PKG_DIR}/DEBIAN/postinst"
+
+          cat > "${PKG_DIR}/DEBIAN/prerm" << 'PRERM'
+          #!/bin/bash
+          set -e
+          if [ -d /run/systemd/system ] && systemctl is-active --quiet rustfs; then
+              systemctl stop rustfs
+          fi
+          PRERM
+          chmod 755 "${PKG_DIR}/DEBIAN/prerm"
+
+          cat > "${PKG_DIR}/DEBIAN/postrm" << 'POSTRM'
+          #!/bin/bash
+          set -e
+          if [ -d /run/systemd/system ]; then
+              systemctl daemon-reload
+          fi
+          POSTRM
+          chmod 755 "${PKG_DIR}/DEBIAN/postrm"
+
+          cp LICENSE "${PKG_DIR}/usr/share/doc/rustfs/"
+          cp README.md "${PKG_DIR}/usr/share/doc/rustfs/"
+
+          fakeroot dpkg-deb --build "${PKG_DIR}"
+          ls -lh "${DEB_FILE}"
+          echo "deb_file=${DEB_FILE}" >> "${GITHUB_OUTPUT}"
+
+      - name: Upload DEB artifact
+        uses: actions/upload-artifact@b7c566a772e6b6bfb58ed0dc250532a479d7789f # v6
+        with:
+          name: ${{ steps.deb.outputs.deb_file }}
+          path: ${{ steps.deb.outputs.deb_file }}
+          if-no-files-found: error
+
+      # Persist the nightly deb on Cloudflare R2 (same channel as package.yml)
+      # so it can be downloaded later with a stable, unauthenticated URL —
+      # e.g. https://dl.rustfs.com/artifacts/rustfs/packages/nightly/... .
+      # Skipped when the R2 secrets are not configured (artifact-only mode).
+      - name: Upload DEB to Cloudflare R2
+        if: env.R2_ACCESS_KEY_ID != ''
+        env:
+          R2_ACCESS_KEY_ID: ${{ secrets.R2_ACCESS_KEY_ID }}
+          R2_SECRET_ACCESS_KEY: ${{ secrets.R2_SECRET_ACCESS_KEY }}
+          R2_ENDPOINT: ${{ secrets.R2_ENDPOINT }}
+          R2_BUCKET: ${{ secrets.R2_BUCKET }}
+          AWS_EC2_METADATA_DISABLED: true
+        shell: bash
+        run: |
+          set -euo pipefail
+
+          if [[ -z "$R2_ACCESS_KEY_ID" || -z "$R2_SECRET_ACCESS_KEY" || -z "$R2_ENDPOINT" || -z "$R2_BUCKET" ]]; then
+            echo "⚠️ R2 credentials missing, skipping upload"
+            exit 0
+          fi
+
+          if ! command -v aws >/dev/null 2>&1; then
+            sudo apt-get update && sudo apt-get install -y -qq awscli
+          fi
+
+          export AWS_ACCESS_KEY_ID="$R2_ACCESS_KEY_ID"
+          export AWS_SECRET_ACCESS_KEY="$R2_SECRET_ACCESS_KEY"
+          export AWS_DEFAULT_REGION="auto"
+
+          DEB_FILE="${{ steps.deb.outputs.deb_file }}"
+          R2_PREFIX="s3://${R2_BUCKET}/artifacts/rustfs/packages/nightly/"
+
+          echo "📤 Uploading ${DEB_FILE} to ${R2_PREFIX}"
+          aws s3 cp "${DEB_FILE}" "${R2_PREFIX}" --endpoint-url "$R2_ENDPOINT" --only-show-errors
+
+          # Stable "latest" alias so tests can fetch the newest nightly
+          # without knowing today's date.
+          echo "📤 Uploading latest alias"
+          aws s3 cp "${DEB_FILE}" "${R2_PREFIX}rustfs-nightly-latest.deb" \
+            --endpoint-url "$R2_ENDPOINT" --only-show-errors
+
+          echo "✅ R2 upload complete"
+
  # Live-Vault lane for the rustfs-kms suite (rustfs/backlog#1774).
  #
  # RUSTFS_KMS_VAULT_TOKEN is the single switch that adds the Vault KV2 and
@@ -27,6 +27,7 @@ on:
    paths:
      - 'flake.nix'
      - 'flake.lock'
+      - 'nix/**'
      - 'Cargo.toml'
      - 'Cargo.lock'
      - '.github/workflows/nix.yml'
@@ -36,6 +37,7 @@ on:
    paths:
      - 'flake.nix'
      - 'flake.lock'
+      - 'nix/**'
      - 'Cargo.toml'
      - 'Cargo.lock'
      - '.github/workflows/nix.yml'
@@ -0,0 +1,110 @@
+# Copyright 2024 RustFS Team
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#     http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+
+name: OIDC Keycloak Live
+
+on:
+  pull_request:
+    paths:
+      - ".github/workflows/oidc-keycloak.yml"
+      - "crates/config/src/constants/oidc.rs"
+      - "crates/iam/src/federation/**"
+      - "crates/iam/src/oidc.rs"
+      - "rustfs/src/admin/handlers/oidc.rs"
+      - "rustfs/src/admin/handlers/sts.rs"
+      - "scripts/test/oidc_keycloak_live.sh"
+      - "scripts/test/fixtures/keycloak-rustfs-ci-realm.json"
+  push:
+    branches: [main]
+    paths:
+      - ".github/workflows/oidc-keycloak.yml"
+      - "crates/config/src/constants/oidc.rs"
+      - "crates/iam/src/federation/**"
+      - "crates/iam/src/oidc.rs"
+      - "rustfs/src/admin/handlers/oidc.rs"
+      - "rustfs/src/admin/handlers/sts.rs"
+      - "scripts/test/oidc_keycloak_live.sh"
+      - "scripts/test/fixtures/keycloak-rustfs-ci-realm.json"
+  schedule:
+    - cron: "23 2 * * 1"
+      timezone: "Asia/Shanghai"
+  workflow_dispatch:
+
+permissions:
+  contents: read
+
+concurrency:
+  group: oidc-keycloak-live-${{ github.ref }}
+  cancel-in-progress: true
+
+jobs:
+  oidc-keycloak-live:
+    name: OIDC Keycloak live gate
+    runs-on: ubuntu-latest
+    timeout-minutes: 60
+    env:
+      FORCE_JAVASCRIPT_ACTIONS_TO_NODE24: "true"
+    steps:
+      - name: Checkout repository
+        uses: actions/checkout@9c091bb21b7c1c1d1991bb908d89e4e9dddfe3e0 # v7
+        with:
+          persist-credentials: false
+
+      - name: Setup Rust environment
+        uses: ./.github/actions/setup
+        with:
+          cache-shared-key: oidc-keycloak-live
+          cache-save-if: "true"
+          install-build-packaging-tools: "false"
+          install-test-tools: "false"
+
+      - name: Build RustFS
+        run: cargo build --locked -p rustfs --bin rustfs
+
+      - name: Install pinned request signer
+        run: |
+          python3 -m pip install --user --upgrade pip "awscurl==0.44"
+          echo "${HOME}/.local/bin" >> "${GITHUB_PATH}"
+
+      - name: Run live Keycloak discovery, JWT and STS checks
+        run: bash scripts/test/oidc_keycloak_live.sh ./target/debug/rustfs
+
+      - name: Upload service logs
+        if: failure()
+        uses: actions/upload-artifact@b7c566a772e6b6bfb58ed0dc250532a479d7789f # v6
+        with:
+          name: oidc-keycloak-live-${{ github.run_number }}
+          path: ${{ runner.temp }}/rustfs-keycloak-live-*/**/*.log
+          if-no-files-found: ignore
+          retention-days: 3
+
+  alert-on-failure:
+    name: Alert on scheduled failure
+    needs: oidc-keycloak-live
+    if: >-
+      always() && github.event_name == 'schedule' &&
+      (needs.oidc-keycloak-live.result == 'failure' || needs.oidc-keycloak-live.result == 'cancelled')
+    runs-on: ubuntu-latest
+    timeout-minutes: 10
+    permissions:
+      contents: read
+      issues: write
+    steps:
+      - uses: actions/checkout@9c091bb21b7c1c1d1991bb908d89e4e9dddfe3e0 # v7
+        with:
+          persist-credentials: false
+      - name: Open or update failure-tracking issue
+        uses: ./.github/actions/schedule-failure-issue
+        with:
+          github-token: ${{ secrets.GITHUB_TOKEN }}
@@ -21,10 +21,10 @@
 # - workflow_run: automatically package after "Build and Release" completes
 #   for a release tag (the mac/windows/linux binaries are already uploaded
 #   to the GitHub release before packaging starts)
-# - workflow_dispatch: manual fallback (backfill / re-run) with optional tag/run_id
+# - workflow_dispatch: manual fallback with a release tag and/or exact build run ID
 #
 # Flow:
-# 1. Resolve the triggering Build workflow run for the release tag
+# 1. Resolve and validate the selected Build workflow run and source identity
 # 2. Download Linux binaries (x86_64-gnu, aarch64-gnu) from build artifacts
 # 3. Build DEB packages for amd64 and arm64
 # 4. Build RPM packages for x86_64 and aarch64
@@ -51,7 +51,7 @@ on:
        required: false
        type: string
      build_run_id:
-        description: "Build workflow run ID (overrides tag lookup)"
+        description: "Build workflow run ID (when combined with tag, both must identify the same release commit)"
        required: false
        type: string

@@ -82,6 +82,9 @@ jobs:
      version: ${{ steps.resolve.outputs.version }}
      build_type: ${{ steps.resolve.outputs.build_type }}
      build_run_id: ${{ steps.resolve.outputs.build_run_id }}
+      build_run_number: ${{ steps.resolve.outputs.build_run_number }}
+      head_sha: ${{ steps.resolve.outputs.head_sha }}
+      dev_sequence: ${{ steps.resolve.outputs.dev_sequence }}
      tag: ${{ steps.resolve.outputs.tag }}
    steps:
      - name: Resolve build run
@@ -89,90 +92,129 @@ jobs:
        shell: bash
        env:
          GH_TOKEN: ${{ github.token }}
+          EVENT_NAME: ${{ github.event_name }}
+          REPOSITORY: ${{ github.repository }}
          INPUT_TAG: ${{ github.event.inputs.tag }}
          INPUT_RUN_ID: ${{ github.event.inputs.build_run_id }}
        run: |
          set -euo pipefail

-          # Determine tag
-          if [[ "${{ github.event_name }}" == "workflow_run" ]]; then
-            TAG="${HEAD_BRANCH}"
-          elif [[ -n "$INPUT_TAG" ]]; then
-            TAG="$INPUT_TAG"
+          fail() {
+            echo "❌ $1" >&2
+            exit 1
+          }
+
+          TAG=""
+          BUILD_RUN_ID=""
+          case "$EVENT_NAME" in
+            workflow_run)
+              TAG="$HEAD_BRANCH"
+              BUILD_RUN_ID="$WORKFLOW_RUN_ID"
+              ;;
+            workflow_dispatch)
+              TAG="$INPUT_TAG"
+              BUILD_RUN_ID="$INPUT_RUN_ID"
+              ;;
+            *) fail "unsupported event: $EVENT_NAME" ;;
+          esac
+
+          # Validate and classify tags before using them in API paths or logs.
+          semver_core='(0|[1-9][0-9]*)\.(0|[1-9][0-9]*)\.(0|[1-9][0-9]*)'
+          prerelease_id='(alpha|beta|rc)\.(0|[1-9][0-9]*)'
+          if [[ -n "$TAG" ]]; then
+            if [[ "$TAG" =~ ^${semver_core}-${prerelease_id}-preview\.(0|[1-9][0-9]*)$ ]]; then
+              BUILD_TYPE=preview
+            elif [[ "$TAG" =~ ^${semver_core}-${prerelease_id}$ ]]; then
+              BUILD_TYPE=prerelease
+            elif [[ "$TAG" =~ ^${semver_core}$ ]]; then
+              BUILD_TYPE=release
+            else
+              fail "tag is not a supported strict package version"
+            fi
          else
-            TAG=""
+            BUILD_TYPE=development
          fi

-          echo "Tag: ${TAG:-<none>}"
-
-          # Determine build run ID
-          BUILD_RUN_ID=""
-
-          if [[ -n "$INPUT_RUN_ID" ]]; then
-            # Explicit run ID takes priority
-            BUILD_RUN_ID="$INPUT_RUN_ID"
-            echo "Using explicit build run ID: $BUILD_RUN_ID"
-
-          elif [[ "${{ github.event_name }}" == "workflow_run" ]]; then
-            # Use the Build and Release run that triggered this workflow
-            BUILD_RUN_ID="${WORKFLOW_RUN_ID}"
-            echo "Using triggering workflow run: $BUILD_RUN_ID"
-
+          if [[ -n "$BUILD_RUN_ID" ]]; then
+            [[ "$BUILD_RUN_ID" =~ ^[1-9][0-9]*$ ]] || fail "build run ID must be a positive decimal integer"
+            echo "Using selected build run: $BUILD_RUN_ID"
          elif [[ -n "$TAG" ]]; then
-            # Find the build run that produced this tag
            echo "Looking for build run for tag: $TAG"
-            BUILD_RUN_ID=$(gh api \
-              "repos/${{ github.repository }}/actions/workflows/build.yml/runs?branch=${TAG}&status=success&per_page=1" \
-              --jq '.workflow_runs[0].id' 2>/dev/null || echo "")
+            BUILD_RUN_ID=$(gh api --method GET \
+              "repos/${REPOSITORY}/actions/workflows/build.yml/runs" \
+              -f branch="$TAG" -f status=success -F per_page=1 \
+              --jq '.workflow_runs[0].id // empty' 2>/dev/null || true)

-            if [[ -z "$BUILD_RUN_ID" || "$BUILD_RUN_ID" == "null" ]]; then
-              # Tag might not be a branch; try event=push with head_branch matching
-              BUILD_RUN_ID=$(gh api \
-                "repos/${{ github.repository }}/actions/workflows/build.yml/runs?event=push&status=success&per_page=100" \
-                --jq ".workflow_runs[] | select(.head_branch == \"$TAG\") | .id" 2>/dev/null | head -1 || echo "")
-            fi
-
-            if [[ -z "$BUILD_RUN_ID" || "$BUILD_RUN_ID" == "null" ]]; then
-              echo "❌ No successful build run found for tag: $TAG"
-              exit 1
+            if [[ -z "$BUILD_RUN_ID" ]]; then
+              BUILD_RUN_ID=$(gh api --method GET \
+                "repos/${REPOSITORY}/actions/workflows/build.yml/runs" \
+                -f event=push -f status=success -F per_page=100 2>/dev/null |
+                jq -r --arg tag "$TAG" \
+                  '[.workflow_runs[] | select(.head_branch == $tag)][0].id // empty' || true)
            fi
+            [[ "$BUILD_RUN_ID" =~ ^[1-9][0-9]*$ ]] || fail "no successful build run found for tag"
            echo "Found build run: $BUILD_RUN_ID"
-
          else
-            # No tag — latest successful main build
            echo "No tag specified, looking for latest main build"
-            BUILD_RUN_ID=$(gh api \
-              "repos/${{ github.repository }}/actions/workflows/build.yml/runs?branch=main&status=success&per_page=1" \
-              --jq '.workflow_runs[0].id' 2>/dev/null || echo "")
-
-            if [[ -z "$BUILD_RUN_ID" || "$BUILD_RUN_ID" == "null" ]]; then
-              echo "❌ No successful main build found"
-              exit 1
-            fi
+            BUILD_RUN_ID=$(gh api --method GET \
+              "repos/${REPOSITORY}/actions/workflows/build.yml/runs" \
+              -f branch=main -f status=success -F per_page=1 \
+              --jq '.workflow_runs[0].id // empty' 2>/dev/null || true)
+            [[ "$BUILD_RUN_ID" =~ ^[1-9][0-9]*$ ]] || fail "no successful main build found"
            echo "Latest main build: $BUILD_RUN_ID"
          fi

-          # Determine version and build type
+          # Fetch once and use the same immutable run metadata for identity,
+          # ordering, workflow provenance, and release-channel validation.
+          RUN_JSON=$(gh api "repos/${REPOSITORY}/actions/runs/${BUILD_RUN_ID}") ||
+            fail "cannot read selected build run"
+          RUN_ID=$(jq -r '.id // empty' <<<"$RUN_JSON")
+          RUN_NUMBER=$(jq -r '.run_number // empty' <<<"$RUN_JSON")
+          RUN_STATUS=$(jq -r '.status // empty' <<<"$RUN_JSON")
+          RUN_CONCLUSION=$(jq -r '.conclusion // empty' <<<"$RUN_JSON")
+          RUN_PATH=$(jq -r '.path // empty' <<<"$RUN_JSON")
+          HEAD_SHA=$(jq -r '.head_sha // empty' <<<"$RUN_JSON")
+          RUN_HEAD_BRANCH=$(jq -r '.head_branch // empty' <<<"$RUN_JSON")
+
+          [[ "$RUN_ID" == "$BUILD_RUN_ID" ]] || fail "run metadata ID mismatch"
+          [[ "$RUN_NUMBER" =~ ^[1-9][0-9]*$ ]] || fail "build run number must be a positive decimal integer"
+          [[ "$RUN_STATUS" == completed && "$RUN_CONCLUSION" == success ]] || fail "selected build run is not successful"
+          [[ "$RUN_PATH" == .github/workflows/build.yml ]] || fail "selected run is not Build and Release"
+          [[ "$HEAD_SHA" =~ ^[0-9a-f]{40}$ ]] || fail "selected build run has an invalid head SHA"
+          [[ "$RUN_HEAD_BRANCH" != *$'\n'* && -n "$RUN_HEAD_BRANCH" ]] || fail "selected build run has an invalid head branch"
+
          if [[ -n "$TAG" ]]; then
+            [[ "$RUN_HEAD_BRANCH" == "$TAG" ]] || fail "tag and build run head branch do not match"
+
+            TAG_REF_JSON=$(gh api "repos/${REPOSITORY}/git/ref/tags/${TAG}") ||
+              fail "cannot resolve release tag ref"
+            TAG_OBJECT_TYPE=$(jq -r '.object.type // empty' <<<"$TAG_REF_JSON")
+            TAG_OBJECT_SHA=$(jq -r '.object.sha // empty' <<<"$TAG_REF_JSON")
+            depth=0
+            while [[ "$TAG_OBJECT_TYPE" == tag && $depth -lt 5 ]]; do
+              TAG_OBJECT_JSON=$(gh api "repos/${REPOSITORY}/git/tags/${TAG_OBJECT_SHA}") ||
+                fail "cannot peel annotated release tag"
+              TAG_OBJECT_TYPE=$(jq -r '.object.type // empty' <<<"$TAG_OBJECT_JSON")
+              TAG_OBJECT_SHA=$(jq -r '.object.sha // empty' <<<"$TAG_OBJECT_JSON")
+              depth=$((depth + 1))
+            done
+            [[ "$TAG_OBJECT_TYPE" == commit && "$TAG_OBJECT_SHA" =~ ^[0-9a-f]{40}$ ]] ||
+              fail "release tag does not resolve to a commit"
+            [[ "$TAG_OBJECT_SHA" == "$HEAD_SHA" ]] || fail "release tag commit and build run head SHA do not match"
            VERSION="$TAG"
-            if [[ "$TAG" == *"-preview"* ]]; then
-              BUILD_TYPE="preview"
-            elif [[ "$TAG" == *"alpha"* || "$TAG" == *"beta"* || "$TAG" == *"rc"* ]]; then
-              BUILD_TYPE="prerelease"
-            else
-              BUILD_TYPE="release"
-            fi
+            DEV_SEQUENCE=""
          else
-            SHORT_SHA=$(gh api "repos/${{ github.repository }}/actions/runs/${BUILD_RUN_ID}" \
-              --jq '.head_sha' 2>/dev/null | head -c 7)
-            VERSION="dev-${SHORT_SHA}"
-            BUILD_TYPE="development"
+            VERSION="dev-${HEAD_SHA}"
+            DEV_SEQUENCE="$RUN_NUMBER"
          fi

          {
            echo "version=$VERSION"
            echo "build_type=$BUILD_TYPE"
            echo "build_run_id=$BUILD_RUN_ID"
+            echo "build_run_number=$RUN_NUMBER"
+            echo "head_sha=$HEAD_SHA"
+            echo "dev_sequence=$DEV_SEQUENCE"
            echo "tag=${TAG}"
          } >> "$GITHUB_OUTPUT"

@@ -180,6 +222,7 @@ jobs:
          echo "  Version: $VERSION"
          echo "  Build type: $BUILD_TYPE"
          echo "  Build run ID: $BUILD_RUN_ID"
+          echo "  Build run number: $RUN_NUMBER"

  # Build DEB and RPM packages for each architecture
  package:
@@ -206,6 +249,22 @@ jobs:
        with:
          persist-credentials: false

+      - name: Normalize package metadata
+        id: versions
+        shell: bash
+        env:
+          BUILD_TYPE: ${{ needs.resolve.outputs.build_type }}
+          SOURCE_VERSION: ${{ needs.resolve.outputs.version }}
+          DEV_SEQUENCE: ${{ needs.resolve.outputs.dev_sequence }}
+          DEB_ARCH: ${{ matrix.deb_arch }}
+          RPM_ARCH: ${{ matrix.rpm_arch }}
+        run: |
+          set -euo pipefail
+
+          normalized=$(./scripts/release/package_versions.sh \
+            "$BUILD_TYPE" "$SOURCE_VERSION" "$DEV_SEQUENCE" "$DEB_ARCH" "$RPM_ARCH")
+          printf '%s\n' "$normalized" >> "$GITHUB_OUTPUT"
+
      - name: Download binary artifact from build run
        uses: actions/download-artifact@37930b1c2abaa49bbe596cd826c3c89aef350131 # v7
        with:
@@ -224,7 +283,7 @@ jobs:
          ZIP_FILE=$(find ./binary-artifact -name "*.zip" -type f | head -1)
          if [[ -z "$ZIP_FILE" ]]; then
            echo "❌ No binary artifact found"
-            ls -la ./binary-artifact/ || true
+            find ./binary-artifact -mindepth 1 -maxdepth 1 -print 2>/dev/null || true
            exit 1
          fi

@@ -239,24 +298,22 @@ jobs:
          fi

          chmod +x ./bin/rustfs
-          ls -lh ./bin/rustfs
+          stat --printf='%n %s bytes\n' ./bin/rustfs
          echo "✅ Binary extracted"

      - name: Build DEB package
        id: deb
        shell: bash
+        env:
+          DEB_VERSION: ${{ steps.versions.outputs.deb_version }}
+          DEB_ARCH: ${{ matrix.deb_arch }}
+          DEB_FILE: ${{ steps.versions.outputs.deb_file }}
        run: |
          set -euo pipefail

-          VERSION="${{ needs.resolve.outputs.version }}"
-          DEB_ARCH="${{ matrix.deb_arch }}"
-          # DEB version: replace - with ~ (1.0.0-beta.12 -> 1.0.0~beta.12)
-          # Use a variable for ~ to prevent tilde expansion by bash
-          TILDE='~'
-          DEB_VERSION="${VERSION/-/$TILDE}"
-          PKG_DIR="rustfs_${DEB_VERSION}_${DEB_ARCH}"
+          PKG_DIR="${DEB_FILE%.deb}"

-          echo "Building DEB: ${PKG_DIR}.deb"
+          echo "Building DEB: ${DEB_FILE}"

          mkdir -p "${PKG_DIR}/DEBIAN"
          mkdir -p "${PKG_DIR}/usr/bin"
@@ -333,26 +390,32 @@ jobs:
          cp LICENSE "${PKG_DIR}/usr/share/doc/rustfs/"
          cp README.md "${PKG_DIR}/usr/share/doc/rustfs/"

-          fakeroot dpkg-deb --build "${PKG_DIR}"
+          fakeroot dpkg-deb --build "${PKG_DIR}" "$DEB_FILE"

-          DEB_FILE="${PKG_DIR}.deb"
-          ls -lh "$DEB_FILE"
+          [[ $(dpkg-deb -f "$DEB_FILE" Package) == rustfs ]]
+          [[ $(dpkg-deb -f "$DEB_FILE" Version) == "$DEB_VERSION" ]]
+          [[ $(dpkg-deb -f "$DEB_FILE" Architecture) == "$DEB_ARCH" ]]
+          dpkg-deb --fsys-tarfile "$DEB_FILE" | tar -tf - | grep -Fx './usr/bin/rustfs' >/dev/null
+          stat --printf='%n %s bytes\n' "$DEB_FILE"
          echo "deb_file=$DEB_FILE" >> "$GITHUB_OUTPUT"
          echo "✅ DEB built: $DEB_FILE"

      - name: Build RPM package
        id: rpm
        shell: bash
+        env:
+          RPM_VERSION: ${{ steps.versions.outputs.rpm_version }}
+          RPM_RELEASE: ${{ steps.versions.outputs.rpm_release }}
+          RPM_ARCH: ${{ matrix.rpm_arch }}
+          RPM_FILE: ${{ steps.versions.outputs.rpm_file }}
        run: |
          set -euo pipefail

-          VERSION="${{ needs.resolve.outputs.version }}"
-          RPM_ARCH="${{ matrix.rpm_arch }}"
-
          echo "Building RPM for ${RPM_ARCH}"

-          sudo apt-get update && sudo apt-get install -y ruby ruby-dev build-essential
+          sudo apt-get update && sudo apt-get install -y ruby ruby-dev build-essential rpm
          sudo gem install fpm
+          ./scripts/test_package_versions.sh --require-package-managers

          # Create config file for fpm (DEB build creates it in its package dir structure,
          # but fpm needs the file to exist before packaging)
@@ -367,8 +430,10 @@ jobs:

          fpm -s dir -t rpm \
            --name rustfs \
-            --version "$VERSION" \
+            --version "$RPM_VERSION" \
+            --iteration "$RPM_RELEASE" \
            --architecture "$RPM_ARCH" \
+            --package "$RPM_FILE" \
            --depends "glibc >= 2.31" \
            --maintainer "RustFS Team <[email protected]>" \
            --description "High-performance distributed object storage" \
@@ -410,13 +475,16 @@ jobs:
            LICENSE=/usr/share/doc/rustfs/LICENSE \
            README.md=/usr/share/doc/rustfs/README.md

-          RPM_FILE=$(ls -1 rustfs-*.rpm 2>/dev/null | head -1)
-          if [[ -z "$RPM_FILE" ]]; then
+          if [[ ! -f "$RPM_FILE" ]]; then
            echo "❌ RPM build failed"
            exit 1
          fi

-          ls -lh "$RPM_FILE"
+          RPM_METADATA=$(rpm -qp --qf '%{NAME}\n%{VERSION}\n%{RELEASE}\n%{ARCH}\n' "$RPM_FILE")
+          EXPECTED_METADATA=$(printf 'rustfs\n%s\n%s\n%s' "$RPM_VERSION" "$RPM_RELEASE" "$RPM_ARCH")
+          [[ "$RPM_METADATA" == "$EXPECTED_METADATA" ]]
+          rpm -qpl "$RPM_FILE" | grep -Fx '/usr/bin/rustfs' >/dev/null
+          stat --printf='%n %s bytes\n' "$RPM_FILE"
          echo "rpm_file=$RPM_FILE" >> "$GITHUB_OUTPUT"
          echo "✅ RPM built: $RPM_FILE"

@@ -437,6 +505,9 @@ jobs:
          R2_ENDPOINT: ${{ secrets.R2_ENDPOINT }}
          R2_BUCKET: ${{ secrets.R2_BUCKET }}
          AWS_EC2_METADATA_DISABLED: true
+          BUILD_TYPE: ${{ needs.resolve.outputs.build_type }}
+          DEB_FILE: ${{ steps.deb.outputs.deb_file }}
+          RPM_FILE: ${{ steps.rpm.outputs.rpm_file }}
        shell: bash
        run: |
          set -euo pipefail
@@ -454,7 +525,6 @@ jobs:
          export AWS_SECRET_ACCESS_KEY="$R2_SECRET_ACCESS_KEY"
          export AWS_DEFAULT_REGION="auto"

-          BUILD_TYPE="${{ needs.resolve.outputs.build_type }}"
          if [[ "$BUILD_TYPE" == "development" ]]; then
            R2_PREFIX="artifacts/rustfs/packages/dev"
          else
@@ -464,9 +534,6 @@ jobs:

          echo "📤 Uploading to $R2_PATH"

-          DEB_FILE="${{ steps.deb.outputs.deb_file }}"
-          RPM_FILE="${{ steps.rpm.outputs.rpm_file }}"
-
          for f in "$DEB_FILE" "$RPM_FILE"; do
            if [[ -n "$f" && -f "$f" ]]; then
              echo "Uploading: $f"
@@ -492,14 +559,13 @@ jobs:
        if: needs.resolve.outputs.tag != ''
        env:
          GH_TOKEN: ${{ github.token }}
+          TAG: ${{ needs.resolve.outputs.tag }}
+          DEB_FILE: ${{ steps.deb.outputs.deb_file }}
+          RPM_FILE: ${{ steps.rpm.outputs.rpm_file }}
        shell: bash
        run: |
          set -euo pipefail

-          TAG="${{ needs.resolve.outputs.tag }}"
-          DEB_FILE="${{ steps.deb.outputs.deb_file }}"
-          RPM_FILE="${{ steps.rpm.outputs.rpm_file }}"
-
          # Upload the packages, then refresh the release checksums so the new
          # assets are covered, matching the binary release flow.
          for f in "$DEB_FILE" "$RPM_FILE"; do
@@ -551,12 +617,19 @@ jobs:
    steps:
      - name: Print summary
        shell: bash
+        env:
+          SUMMARY_VERSION: ${{ needs.resolve.outputs.version }}
+          SUMMARY_BUILD_TYPE: ${{ needs.resolve.outputs.build_type }}
+          SUMMARY_BUILD_RUN_ID: ${{ needs.resolve.outputs.build_run_id }}
+          SUMMARY_PACKAGE_STATUS: ${{ needs.package.result }}
        run: |
-          echo "## 📦 Package Summary" >> "$GITHUB_STEP_SUMMARY"
-          echo "" >> "$GITHUB_STEP_SUMMARY"
-          echo "| Item | Value |" >> "$GITHUB_STEP_SUMMARY"
-          echo "|------|-------|" >> "$GITHUB_STEP_SUMMARY"
-          echo "| Version | \`${{ needs.resolve.outputs.version }}\` |" >> "$GITHUB_STEP_SUMMARY"
-          echo "| Build Type | ${{ needs.resolve.outputs.build_type }} |" >> "$GITHUB_STEP_SUMMARY"
-          echo "| Build Run | #${{ needs.resolve.outputs.build_run_id }} |" >> "$GITHUB_STEP_SUMMARY"
-          echo "| Package Status | ${{ needs.package.result }} |" >> "$GITHUB_STEP_SUMMARY"
+          {
+            echo "## 📦 Package Summary"
+            echo ""
+            echo "| Item | Value |"
+            echo "|------|-------|"
+            echo "| Version | \`${SUMMARY_VERSION}\` |"
+            echo "| Build Type | ${SUMMARY_BUILD_TYPE} |"
+            echo "| Build Run | #${SUMMARY_BUILD_RUN_ID} |"
+            echo "| Package Status | ${SUMMARY_PACKAGE_STATUS} |"
+          } >> "$GITHUB_STEP_SUMMARY"
@@ -0,0 +1,74 @@
+# Copyright 2024 RustFS Team
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#     http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+
+# Functional chain driver: runs the ten functional suites in a fixed order
+# (upgrade -> s3 -> kms -> tier -> storage -> heal -> pool -> security ->
+# replication, with performance on its own runner in parallel) and guarantees
+# the chain keeps moving even when individual suites fail.
+#
+# Each suite workflow can still be dispatched standalone (workflow_dispatch);
+# only chain-triggered runs forward to the next suite via repository_dispatch,
+# so a standalone run never drags the rest of the chain behind it.
+#
+# Why not workflow_run chaining: GitHub does not guarantee delivery of
+# workflow_run events (they are fire-and-forget), and the head-SHA filter made
+# newly added suites (storage) unable to trigger at all. Explicit
+# repository_dispatch handoffs are verifiable and re-drivable.
+
+name: RustFS Functional Chain
+
+on:
+  workflow_dispatch:
+  workflow_run:
+    # Entry point: start the chain after the nightly build completes. The
+    # build's own conclusion does not gate the chain; each suite reports its
+    # own result to rustfs/backlog and the dashboard.
+    workflows: ["Nightly GNU Build"]
+    types: [completed]
+
+permissions:
+  contents: read
+
+jobs:
+  start-chain:
+    name: Start functional chain (upgrade first)
+    runs-on: ubuntu-latest
+    timeout-minutes: 10
+    if: ${{ github.event_name == 'workflow_dispatch' || (github.event_name == 'workflow_run' && github.event.workflow_run.event == 'schedule') }}
+    steps:
+      - name: Dispatch first suite (upgrade)
+        env:
+          GH_TOKEN: ${{ secrets.PF_TESTING_GH_TOKEN }}
+        run: |
+          set -euo pipefail
+          if [ -z "${GH_TOKEN:-}" ]; then
+            echo "PF_TESTING_GH_TOKEN is not configured; cannot start the functional chain" >&2
+            exit 1
+          fi
+          gh api --method POST repos/rustfs/rustfs/dispatches \
+            -f event_type='rustfs-chain-upgrade' \
+            -F 'client_payload[from_suite]=nightly-build'
+
+      - name: Dispatch performance suite (parallel, own runner)
+        env:
+          GH_TOKEN: ${{ secrets.PF_TESTING_GH_TOKEN }}
+        run: |
+          set -euo pipefail
+          if [ -z "${GH_TOKEN:-}" ]; then
+            echo "PF_TESTING_GH_TOKEN is not configured; cannot dispatch performance" >&2
+            exit 1
+          fi
+          gh api --method POST repos/rustfs/rustfs/dispatches \
+            -f event_type='rustfs-chain-performance' \
+            -F 'client_payload[from_suite]=nightly-build'
@@ -0,0 +1,339 @@
+name: RustFS Heal Test
+
+on:
+  workflow_dispatch:
+    inputs:
+      package_url:
+        description: 'Direct .deb URL (nightly/R2). Defaults to the latest nightly deb.'
+        required: false
+        type: string
+      stop_node_gb:
+        description: 'Stop the outage node when surviving nodes reach N GiB'
+        required: false
+        default: '15'
+      warp_stop_gb:
+        description: 'Stop warp when surviving nodes reach N GiB'
+        required: false
+        default: '40'
+      cleanup_before:
+        description: 'Reset the nodes before the test (DESTROYS existing data/config)'
+        type: boolean
+        default: true
+      cleanup_after:
+        description: 'Reset the nodes after the test (DESTROYS test data/config)'
+        type: boolean
+        default: true
+  repository_dispatch:
+    # Chain handoff: dispatched when the storage suite finishes. Heal runs
+    # exactly once per chain; the pool expansion workflow no longer embeds
+    # its own heal pass.
+    types: [rustfs-chain-heal]
+
+permissions:
+  contents: read
+
+# Only one test at a time: both this and the pool-expansion workflow mutate
+# the same test environment, so they share one concurrency group.
+concurrency:
+  group: rustfs-shared-functional-tests
+  cancel-in-progress: false
+
+defaults:
+  run:
+    shell: bash
+
+env:
+  RUSTFS_ACCESS_KEY: ${{ secrets.RUSTFS_ACCESS_KEY }}
+  RUSTFS_SECRET_KEY: ${{ secrets.RUSTFS_SECRET_KEY }}
+  RUSTFS_API_ENDPOINT: ${{ secrets.RUSTFS_API_ENDPOINT || vars.RUSTFS_API_ENDPOINT || vars.RUSTFS_RC_ENDPOINT }}
+  RUSTFS_NODES: ${{ secrets.RUSTFS_NODES || vars.RUSTFS_NODES }}
+  RUSTFS_SSH_USER: ${{ secrets.RUSTFS_SSH_USER || vars.RUSTFS_SSH_USER }}
+  PF_TESTING_GH_TOKEN: ${{ secrets.PF_TESTING_GH_TOKEN }}
+  RUSTFS_NIGHTLY_PACKAGE_URL: ${{ vars.RUSTFS_NIGHTLY_PACKAGE_URL || 'https://dl.rustfs.com/artifacts/rustfs/packages/nightly/rustfs-nightly-latest.deb' }}
+
+jobs:
+  heal-test:
+    runs-on: smoke-testing
+    # Requirement: a failing suite must not fail the workflow; failures
+    # are filed to rustfs/backlog and the chain continues.
+    continue-on-error: true
+    timeout-minutes: 480
+    # Standalone manual run, or one link of the nightly functional chain
+    # (storage -> heal -> pool). Pool expansion no longer re-runs heal.
+    if: ${{ github.event_name == 'workflow_dispatch' || github.event_name == 'repository_dispatch' }}
+    steps:
+      # auto-testing is private: clone it with the dedicated PF token (not
+      # GITHUB_TOKEN) and retry transient GitHub/network failures.
+      - name: Checkout auto-testing scripts (with retry)
+        env:
+          GH_TOKEN: ${{ secrets.PF_TESTING_GH_TOKEN }}
+        run: |
+          set -euo pipefail
+          rm -rf auto-testing
+          for attempt in 1 2 3 4 5; do
+            if gh repo clone rustfs/auto-testing auto-testing -- --depth 1 --quiet; then
+              echo "auto-testing cloned (attempt ${attempt})"
+              exit 0
+            fi
+            rm -rf auto-testing
+            echo "clone attempt ${attempt} failed; retrying in $((attempt * 15))s" >&2
+            sleep $((attempt * 15))
+          done
+          echo "ERROR: unable to clone rustfs/auto-testing after 5 attempts" >&2
+          exit 1
+
+      - name: Show environment
+        run: |
+          uname -a
+          jq --version
+          openssl version
+          warp --version || true
+          df -h /data | tail -1
+
+      - name: Cleanup environment (before)
+        if: ${{ inputs.cleanup_before != 'false' }}
+        run: |
+          set -euo pipefail
+          read -r -a NODES <<< "${RUSTFS_NODES:-vm000 vm001 vm002}"
+          SSH_USER="${RUSTFS_SSH_USER:-azureuser}"
+          for node in "${NODES[@]}"; do
+            ssh -o BatchMode=yes -o ConnectTimeout=10 -o StrictHostKeyChecking=accept-new "${SSH_USER}@${node}" '
+              set -euo pipefail
+              SUDO=""; [ "$(id -u)" -ne 0 ] && SUDO="sudo -n"
+              ${SUDO} systemctl stop rustfs 2>/dev/null || true
+              if ${SUDO} dpkg -l rustfs 2>/dev/null | grep -q "^ii"; then
+                ${SUDO} dpkg -P rustfs
+              fi
+              for i in 1 2 3 4; do ${SUDO} rm -rf /data/rustfs${i}/mnmd; done
+              ${SUDO} rm -rf /var/log/rustfs /var/lib/rustfs/kms /var/lib/rustfs/kms-backup
+            '
+          done
+
+      - name: Install RustFS package & start cluster
+        run: |
+          ARGS=(--steps "1,2" -y --endpoint "${{ env.RUSTFS_API_ENDPOINT }}")
+          if [ -n "${{ inputs.package_url }}" ]; then
+            ARGS+=(--package-url "${{ inputs.package_url }}")
+          else
+            ARGS+=(--package-url "${{ env.RUSTFS_NIGHTLY_PACKAGE_URL }}")
+          fi
+          ./auto-testing/rustfs_heal_test.sh "${ARGS[@]}"
+
+      - name: Preflight checks
+        run: |
+          ARGS=(--preflight --endpoint "${{ env.RUSTFS_API_ENDPOINT }}")
+          if [ -n "${{ inputs.package_url }}" ]; then
+            ARGS+=(--package-url "${{ inputs.package_url }}")
+          else
+            ARGS+=(--package-url "${{ env.RUSTFS_NIGHTLY_PACKAGE_URL }}")
+          fi
+          ./auto-testing/rustfs_heal_test.sh "${ARGS[@]}"
+
+      - name: Run heal test (write -> outage -> heal -> verify)
+        id: test
+        run: |
+          ./auto-testing/rustfs_heal_test.sh \
+            --steps "3,4,5,6,7" -y \
+            --endpoint "${{ env.RUSTFS_API_ENDPOINT }}" \
+            --stop-node-gb "${{ inputs.stop_node_gb || '15' }}" \
+            --warp-stop-gb "${{ inputs.warp_stop_gb || '40' }}" \
+            --log-file /tmp/rustfs-heal-test.log
+
+      - name: Generate report
+        if: always()
+        env:
+          LOG_FILE: /tmp/rustfs-heal-test.log
+          REPORT_FILE: /tmp/rustfs-heal-report.md
+        run: |
+          set -euo pipefail
+          PACKAGE_URL='${{ inputs.package_url }}'
+          if [ -n "${PACKAGE_URL}" ]; then
+            PACKAGE_SOURCE="${PACKAGE_URL}"
+          else
+            PACKAGE_SOURCE="${RUSTFS_NIGHTLY_PACKAGE_URL}"
+          fi
+          {
+            echo "# RustFS heal test report"
+            echo ""
+            echo "- Run: ${{ github.server_url }}/${{ github.repository }}/actions/runs/${{ github.run_id }}"
+            echo "- Trigger: ${{ github.event_name }}"
+            echo "- Package: ${PACKAGE_SOURCE}"
+            echo "- Test Step Outcome: ${{ steps.test.outcome }}"
+            echo ""
+            echo "## Log tail"
+            echo '```text'
+            tail -n 200 "${LOG_FILE}" || true
+            echo '```'
+          } | tee "${REPORT_FILE}"
+          cat "${REPORT_FILE}" >> "${GITHUB_STEP_SUMMARY}"
+
+      - name: Upload functional report to dashboard
+        if: always()
+        continue-on-error: true
+        env:
+          GH_TOKEN: ${{ env.PF_TESTING_GH_TOKEN }}
+          REPORT_FILE: /tmp/rustfs-heal-report.md
+          SUITE: heal
+        run: |
+          set -euo pipefail
+          if [ -z "${GH_TOKEN:-}" ]; then
+            echo "PF_TESTING_GH_TOKEN is not configured; skipping dashboard upload"
+            exit 0
+          fi
+          DATE="$(date -u +%Y-%m-%d)"
+          REPORT_PATH="functional-reports/${SUITE}/${DATE}.md"
+          CONTENT="$(python3 -c 'import base64,sys;print(base64.b64encode(open(sys.argv[1],"rb").read()).decode())' "${REPORT_FILE}")"
+          SHA="$(gh api "repos/rustfs/dashboard/contents/${REPORT_PATH}" -q '.sha' 2>/dev/null || true)"
+          if [ -n "${SHA}" ]; then
+            jq -n --arg msg "report(${SUITE}): ${DATE}" --arg content "${CONTENT}" --arg sha "${SHA}" \
+              '{message:$msg, content:$content, sha:$sha}' \
+              | gh api --method PUT "repos/rustfs/dashboard/contents/${REPORT_PATH}" --input - >/dev/null
+          else
+            jq -n --arg msg "report(${SUITE}): ${DATE}" --arg content "${CONTENT}" \
+              '{message:$msg, content:$content}' \
+              | gh api --method PUT "repos/rustfs/dashboard/contents/${REPORT_PATH}" --input - >/dev/null
+          fi
+
+      - name: File failure issue in rustfs/backlog
+        if: ${{ always() && (failure() || steps.test.outcome == 'failure' || steps.test.outcome == 'cancelled') }}
+        continue-on-error: true
+        env:
+          GH_TOKEN: ${{ secrets.PF_TESTING_GH_TOKEN }}
+          SUITE: 'heal'
+          SUITE_LABEL: 'Heal'
+          RUN_URL: ${{ github.server_url }}/${{ github.repository }}/actions/runs/${{ github.run_id }}
+          REPORT_FILE: '/tmp/rustfs-heal-report.md'
+          LOG_FILE: '/tmp/rustfs-heal-test.log'
+        run: |
+          set -euo pipefail
+          if [ -z "${GH_TOKEN:-}" ]; then
+            echo "PF_TESTING_GH_TOKEN is not configured; skipping backlog issue"
+            exit 0
+          fi
+          TITLE="[functional][${SUITE}] ${SUITE_LABEL} suite failed (run ${GITHUB_RUN_ID})"
+          EXISTING="$(gh issue list -R rustfs/backlog --state all \
+            --search "in:title \"run ${GITHUB_RUN_ID}\"" \
+            --json number --jq '.[].number' || true)"
+          if [ -n "${EXISTING}" ]; then
+            echo "backlog issue already exists for run ${GITHUB_RUN_ID}; skipping"
+            exit 0
+          fi
+          redact() {
+            sed -E \
+              -e 's/(RUSTFS_(ACCESS_KEY|SECRET_KEY)[=: ]+)[^[:space:]]+/\1[REDACTED]/Ig' \
+              -e 's/(Authorization:).*/\1 [REDACTED]/Ig' \
+              -e 's/(X-Amz-Signature=)[^&[:space:]]+/\1[REDACTED]/Ig' \
+              -e 's/^.*(password|secret|token)[=: ].*/[REDACTED SENSITIVE LINE]/Ig'
+          }
+          BODY_FILE="$(mktemp)"
+          {
+            echo "The **${SUITE_LABEL}** functional suite failed."
+            echo ""
+            echo "- Suite: \`${SUITE}\`"
+            echo "- Run: ${RUN_URL}"
+            echo "- Trigger: ${GITHUB_EVENT_NAME}"
+            echo "- Date: $(date -u +%Y-%m-%d)"
+            echo ""
+            echo "## Report (errors and symptoms)"
+            echo ""
+            if [ -s "${REPORT_FILE}" ]; then
+              redact < "${REPORT_FILE}"
+            elif [ -s "${LOG_FILE:-}" ]; then
+              echo "(report file missing; log tail below)"
+              echo ""
+              tail -n 200 "${LOG_FILE}" | redact
+            else
+              echo "(no report or log file was produced)"
+            fi
+          } | head -c 55000 > "${BODY_FILE}"
+          gh label create functional-test -R rustfs/backlog --color d73a4a 2>/dev/null || true
+          if ! gh issue create -R rustfs/backlog --title "${TITLE}" \
+              --body-file "${BODY_FILE}" --label functional-test; then
+            gh issue create -R rustfs/backlog --title "${TITLE}" --body-file "${BODY_FILE}"
+          fi
+          echo "filed backlog issue for suite ${SUITE}"
+
+      - name: Upload test logs
+        if: always()
+        uses: actions/upload-artifact@b7c566a772e6b6bfb58ed0dc250532a479d7789f # v6
+        with:
+          name: rustfs-heal-test-${{ github.run_id }}
+          path: |
+            /tmp/rustfs-heal-test*.log
+            /tmp/rustfs-warp.*.log
+          if-no-files-found: warn
+
+      - name: Cleanup environment (after)
+        if: ${{ always() && inputs.cleanup_after != 'false' }}
+        run: |
+          set -euo pipefail
+          read -r -a NODES <<< "${RUSTFS_NODES:-vm000 vm001 vm002}"
+          SSH_USER="${RUSTFS_SSH_USER:-azureuser}"
+          for node in "${NODES[@]}"; do
+            ssh -o BatchMode=yes -o ConnectTimeout=10 -o StrictHostKeyChecking=accept-new "${SSH_USER}@${node}" '
+              set -euo pipefail
+              SUDO=""; [ "$(id -u)" -ne 0 ] && SUDO="sudo -n"
+              ${SUDO} systemctl stop rustfs 2>/dev/null || true
+              if ${SUDO} dpkg -l rustfs 2>/dev/null | grep -q "^ii"; then
+                ${SUDO} dpkg -P rustfs
+              fi
+              for i in 1 2 3 4; do ${SUDO} rm -rf /data/rustfs${i}/mnmd; done
+              ${SUDO} rm -rf /var/log/rustfs /var/lib/rustfs/kms /var/lib/rustfs/kms-backup
+            '
+          done
+
+      - name: "Continue functional chain (next: Pool expansion)"
+        # Only chain-triggered runs forward to the next suite; standalone
+        # workflow_dispatch runs stop after their own cleanup. A failed
+        # handoff must never pass silently: it retries, then files an alert
+        # issue in rustfs/backlog so a stalled chain is visible.
+        if: ${{ always() && github.event_name == 'repository_dispatch' }}
+        continue-on-error: true
+        env:
+          GH_TOKEN: ${{ secrets.PF_TESTING_GH_TOKEN }}
+        run: |
+          set -uo pipefail
+          if [ -z "${GH_TOKEN:-}" ]; then
+            echo "PF_TESTING_GH_TOKEN is not configured; cannot dispatch the next suite" >&2
+            exit 1
+          fi
+          DISPATCHED=0
+          for attempt in 1 2 3; do
+            if gh api --method POST repos/rustfs/rustfs/dispatches \
+              -f event_type='rustfs-chain-pool' \
+              -F 'client_payload[from_suite]=heal'; then
+              echo "dispatched next suite Pool expansion (attempt ${attempt})"
+              DISPATCHED=1
+              break
+            fi
+            echo "dispatch attempt ${attempt} failed; retrying in ${attempt}0s" >&2
+            sleep "${attempt}0"
+          done
+          if [ "${DISPATCHED:-0}" -ne 1 ]; then
+            echo "ERROR: functional chain stalled: could not dispatch Pool expansion after 3 attempts" >&2
+            TITLE="[functional][chain] stalled after heal (run ${GITHUB_RUN_ID})"
+            BODY_FILE="$(mktemp)"
+            {
+              echo "The functional chain could not hand off from **heal** to **Pool expansion** after 3 attempts."
+              echo ""
+              echo "- Failed suite job: ${GITHUB_SERVER_URL}/${GITHUB_REPOSITORY}/actions/runs/${GITHUB_RUN_ID}"
+              echo "- Expected next event: 'rustfs-chain-pool'"
+              echo "- Likely cause: PF_TESTING_GH_TOKEN lacks contents:write on rustfs/rustfs, or the GitHub API was unavailable."
+              echo "- Recovery: re-dispatch manually with"
+              FENCE="$(printf "\x60\x60\x60")"; echo "  ${FENCE}"
+              echo "  gh api --method POST repos/rustfs/rustfs/dispatches -f event_type='rustfs-chain-pool'"
+              FENCE="$(printf "\x60\x60\x60")"; echo "  ${FENCE}"
+            } > "${BODY_FILE}"
+            gh issue create -R rustfs/backlog --title "${TITLE}" \
+              --body-file "${BODY_FILE}" --label functional-test \
+              || gh issue create -R rustfs/backlog --title "${TITLE}" --body-file "${BODY_FILE}" \
+              || echo "could not file the stall alert issue either; check the token" >&2
+            exit 1
+          fi
+
+      - name: Notify on failure
+        if: failure()
+        run: |
+          echo "RustFS heal test failed"
+          echo "Package source: ${{ inputs.package_url || 'nightly (R2 latest)' }}"
+          echo "See the uploaded log artifact for details."
@@ -0,0 +1,394 @@
+name: RustFS KMS Test
+
+on:
+  workflow_dispatch:
+    inputs:
+      rustfs_version:
+        description: 'RustFS release tag to test (e.g. 1.0.0-rc.4-preview.1)'
+        required: false
+        default: '1.0.0-rc.4-preview.1'
+      package_url:
+        description: 'Direct .deb URL (nightly/R2/dev). Overrides rustfs_version.'
+        required: false
+        type: string
+      enforce_sse_key_policy:
+        description: 'Enable RUSTFS_KMS_ENFORCE_SSE_KEY_POLICY (runs KMS-401/402)'
+        type: boolean
+        default: false
+      frame_v2:
+        description: 'Enable RUSTFS_ENCRYPTION_FRAME_V2 (runs KMS-318)'
+        type: boolean
+        default: false
+      config_secret:
+        description: 'Set RUSTFS_KMS_CONFIG_SECRET (runs KMS-107 config sealing)'
+        required: false
+        type: string
+  repository_dispatch:
+    # Chain handoff: dispatched when the S3 compatibility suite finishes.
+    types: [rustfs-chain-kms]
+
+permissions:
+  contents: read
+
+concurrency:
+  group: rustfs-shared-functional-tests
+  cancel-in-progress: false
+
+defaults:
+  run:
+    shell: bash
+
+env:
+  RUSTFS_ACCESS_KEY: ${{ secrets.RUSTFS_ACCESS_KEY }}
+  RUSTFS_SECRET_KEY: ${{ secrets.RUSTFS_SECRET_KEY }}
+  RUSTFS_NODES: ${{ secrets.RUSTFS_NODES || vars.RUSTFS_NODES }}
+  RUSTFS_SSH_USER: ${{ secrets.RUSTFS_SSH_USER || vars.RUSTFS_SSH_USER }}
+  RUSTFS_NIGHTLY_PACKAGE_URL: ${{ vars.RUSTFS_NIGHTLY_PACKAGE_URL || 'https://dl.rustfs.com/artifacts/rustfs/packages/nightly/rustfs-nightly-latest.deb' }}
+  PF_TESTING_GH_TOKEN: ${{ secrets.PF_TESTING_GH_TOKEN }}
+
+jobs:
+  kms-test:
+    runs-on: smoke-testing
+    continue-on-error: true
+    timeout-minutes: 420
+    if: ${{ github.event_name == 'workflow_dispatch' || github.event_name == 'repository_dispatch' }}
+    steps:
+      # auto-testing is private: clone it with the dedicated PF token (not
+      # GITHUB_TOKEN) and retry transient GitHub/network failures.
+      - name: Checkout auto-testing scripts (with retry)
+        env:
+          GH_TOKEN: ${{ secrets.PF_TESTING_GH_TOKEN }}
+        run: |
+          set -euo pipefail
+          rm -rf auto-testing
+          for attempt in 1 2 3 4 5; do
+            if gh repo clone rustfs/auto-testing auto-testing -- --depth 1 --quiet; then
+              echo "auto-testing cloned (attempt ${attempt})"
+              exit 0
+            fi
+            rm -rf auto-testing
+            echo "clone attempt ${attempt} failed; retrying in $((attempt * 15))s" >&2
+            sleep $((attempt * 15))
+          done
+          echo "ERROR: unable to clone rustfs/auto-testing after 5 attempts" >&2
+          exit 1
+
+      - name: Show environment
+        run: |
+          uname -a
+          jq --version
+          openssl version
+          docker --version || true
+
+      - name: Cleanup environment (before)
+        run: |
+          set -euo pipefail
+          read -r -a NODES <<< "${RUSTFS_NODES:-vm000 vm001 vm002}"
+          SSH_USER="${RUSTFS_SSH_USER:-azureuser}"
+          for node in "${NODES[@]}"; do
+            ssh -o BatchMode=yes -o ConnectTimeout=10 -o StrictHostKeyChecking=accept-new "${SSH_USER}@${node}" '
+              set -euo pipefail
+              SUDO=""; [ "$(id -u)" -ne 0 ] && SUDO="sudo -n"
+              ${SUDO} systemctl stop rustfs 2>/dev/null || true
+              if ${SUDO} dpkg -l rustfs 2>/dev/null | grep -q "^ii"; then
+                ${SUDO} dpkg -P rustfs
+              fi
+              for i in 1 2 3 4; do ${SUDO} rm -rf /data/rustfs${i}/mnmd; done
+              ${SUDO} rm -rf /var/log/rustfs /var/lib/rustfs/kms /var/lib/rustfs/kms-backup
+            '
+          done
+
+      - name: Ensure docker (Vault container)
+        run: |
+          if ! command -v docker >/dev/null 2>&1; then
+            sudo apt-get update
+            sudo apt-get install -y docker.io
+          fi
+          sudo systemctl enable --now docker
+          docker info >/dev/null 2>&1 || sudo docker info >/dev/null 2>&1
+
+      - name: Run KMS suite
+        id: test
+        continue-on-error: true
+        env:
+          LOG_FILE: /tmp/rustfs-kms.log
+        run: |
+          set -euo pipefail
+          chmod +x auto-testing/rustfs-kms-test.sh
+          PACKAGE_URL='${{ inputs.package_url }}'
+          RUSTFS_VERSION='${{ inputs.rustfs_version }}'
+          ARGS=(--all-topologies --backends "local,vault-kv2" -y --log-file "${LOG_FILE}")
+          EXTRA_ENV=""
+          if [ "${{ inputs.enforce_sse_key_policy }}" = "true" ]; then
+            EXTRA_ENV+="RUSTFS_KMS_ENFORCE_SSE_KEY_POLICY=true"$'\n'
+          fi
+          if [ "${{ inputs.frame_v2 }}" = "true" ]; then
+            EXTRA_ENV+="RUSTFS_ENCRYPTION_FRAME_V2=true"$'\n'
+          fi
+          if [ -n "${{ inputs.config_secret }}" ]; then
+            EXTRA_ENV+="RUSTFS_KMS_CONFIG_SECRET=${{ inputs.config_secret }}"$'\n'
+          fi
+          if [ -n "${EXTRA_ENV}" ]; then
+            ARGS+=(--extra-env "${EXTRA_ENV}")
+          fi
+          if [ -n "${PACKAGE_URL}" ]; then
+            ARGS+=(--package-url "${PACKAGE_URL}")
+          elif [ -n "${RUSTFS_VERSION}" ]; then
+            ARGS+=(--version "${RUSTFS_VERSION}")
+          else
+            ARGS+=(--package-url "${RUSTFS_NIGHTLY_PACKAGE_URL}")
+          fi
+          ./auto-testing/rustfs-kms-test.sh "${ARGS[@]}"
+
+      - name: Generate report
+        if: always()
+        env:
+          LOG_FILE: /tmp/rustfs-kms.log
+          REPORT_FILE: /tmp/rustfs-kms-report.md
+        run: |
+          set -euo pipefail
+          PACKAGE_URL='${{ inputs.package_url }}'
+          RUSTFS_VERSION='${{ inputs.rustfs_version }}'
+          if [ -n "${PACKAGE_URL}" ]; then
+            PACKAGE_SOURCE="${PACKAGE_URL}"
+          elif [ -n "${RUSTFS_VERSION}" ]; then
+            PACKAGE_SOURCE="version ${RUSTFS_VERSION}"
+          else
+            PACKAGE_SOURCE="${RUSTFS_NIGHTLY_PACKAGE_URL}"
+          fi
+          CASE_TABLE="/tmp/rustfs-kms-cases.md"
+          python3 - "${LOG_FILE}" "${CASE_TABLE}" <<'PY'
+          import re
+          import sys
+
+          log_file, out_file = sys.argv[1], sys.argv[2]
+          ansi = re.compile(r'\x1b\[[0-9;]*m')
+          start_re = re.compile(r'^---\s+([A-Z]+-[0-9]+)\s+(.+?)\s+---$')
+          done_re = re.compile(r'^\[(PASS|FAIL|UNSUPPORTED)\]\s+([A-Z]+-[0-9]+)\b')
+
+          rows = []
+          index = {}
+          try:
+            with open(log_file, 'r', encoding='utf-8', errors='replace') as fh:
+              for raw in fh:
+                line = ansi.sub('', raw).strip()
+                m = start_re.match(line)
+                if m:
+                  case_id, name = m.group(1), m.group(2)
+                  if case_id not in index:
+                    index[case_id] = len(rows)
+                    rows.append([case_id, name, 'RUNNING'])
+                  continue
+                m = done_re.match(line)
+                if m:
+                  status, case_id = m.group(1), m.group(2)
+                  if case_id in index:
+                    rows[index[case_id]][2] = status
+                  else:
+                    rows.append([case_id, case_id, status])
+                    index[case_id] = len(rows) - 1
+          except FileNotFoundError:
+            rows = []
+
+          counts = {'PASS': 0, 'FAIL': 0, 'UNSUPPORTED': 0, 'RUNNING': 0}
+          for _, _, status in rows:
+            counts[status] = counts.get(status, 0) + 1
+
+          with open(out_file, 'w', encoding='utf-8') as out:
+            out.write('## Case Summary\n\n')
+            out.write(f"- Total: {len(rows)}\\n")
+            out.write(f"- PASS: {counts.get('PASS', 0)}\\n")
+            out.write(f"- FAIL: {counts.get('FAIL', 0)}\\n")
+            out.write(f"- UNSUPPORTED: {counts.get('UNSUPPORTED', 0)}\\n")
+            out.write('\\n')
+            out.write('| Case | Name | Status |\\n')
+            out.write('| --- | --- | --- |\\n')
+            for case_id, name, status in rows:
+              out.write(f'| {case_id} | {name} | {status} |\\n')
+          PY
+          {
+            echo "# RustFS KMS test report"
+            echo ""
+            echo "- Run: ${{ github.server_url }}/${{ github.repository }}/actions/runs/${{ github.run_id }}"
+            echo "- Trigger: ${{ github.event_name }}"
+            echo "- Package: ${PACKAGE_SOURCE}"
+            echo "- Test Step Outcome: ${{ steps.test.outcome }}"
+            echo ""
+            cat "${CASE_TABLE}" || true
+            echo ""
+            echo "## Log tail"
+            echo '```text'
+            tail -n 200 "${LOG_FILE}" || true
+            echo '```'
+          } | tee "${REPORT_FILE}"
+          cat "${REPORT_FILE}" >> "${GITHUB_STEP_SUMMARY}"
+
+      - name: Upload functional report to dashboard
+        if: always()
+        continue-on-error: true
+        env:
+          GH_TOKEN: ${{ env.PF_TESTING_GH_TOKEN }}
+          REPORT_FILE: /tmp/rustfs-kms-report.md
+          SUITE: kms
+        run: |
+          set -euo pipefail
+          if [ -z "${GH_TOKEN:-}" ]; then
+            echo "PF_TESTING_GH_TOKEN is not configured; skipping dashboard upload"
+            exit 0
+          fi
+          DATE="$(date -u +%Y-%m-%d)"
+          REPORT_PATH="functional-reports/${SUITE}/${DATE}.md"
+          CONTENT="$(python3 -c 'import base64,sys;print(base64.b64encode(open(sys.argv[1],"rb").read()).decode())' "${REPORT_FILE}")"
+          SHA="$(gh api "repos/rustfs/dashboard/contents/${REPORT_PATH}" -q '.sha' 2>/dev/null || true)"
+          if [ -n "${SHA}" ]; then
+            jq -n --arg msg "report(${SUITE}): ${DATE}" --arg content "${CONTENT}" --arg sha "${SHA}" \
+              '{message:$msg, content:$content, sha:$sha}' \
+              | gh api --method PUT "repos/rustfs/dashboard/contents/${REPORT_PATH}" --input - >/dev/null
+          else
+            jq -n --arg msg "report(${SUITE}): ${DATE}" --arg content "${CONTENT}" \
+              '{message:$msg, content:$content}' \
+              | gh api --method PUT "repos/rustfs/dashboard/contents/${REPORT_PATH}" --input - >/dev/null
+          fi
+
+      - name: File failure issue in rustfs/backlog
+        if: ${{ always() && (failure() || steps.test.outcome == 'failure' || steps.test.outcome == 'cancelled') }}
+        continue-on-error: true
+        env:
+          GH_TOKEN: ${{ secrets.PF_TESTING_GH_TOKEN }}
+          SUITE: 'kms'
+          SUITE_LABEL: 'KMS'
+          RUN_URL: ${{ github.server_url }}/${{ github.repository }}/actions/runs/${{ github.run_id }}
+          REPORT_FILE: '/tmp/rustfs-kms-report.md'
+          LOG_FILE: '/tmp/rustfs-kms.log'
+        run: |
+          set -euo pipefail
+          if [ -z "${GH_TOKEN:-}" ]; then
+            echo "PF_TESTING_GH_TOKEN is not configured; skipping backlog issue"
+            exit 0
+          fi
+          TITLE="[functional][${SUITE}] ${SUITE_LABEL} suite failed (run ${GITHUB_RUN_ID})"
+          EXISTING="$(gh issue list -R rustfs/backlog --state all \
+            --search "in:title \"run ${GITHUB_RUN_ID}\"" \
+            --json number --jq '.[].number' || true)"
+          if [ -n "${EXISTING}" ]; then
+            echo "backlog issue already exists for run ${GITHUB_RUN_ID}; skipping"
+            exit 0
+          fi
+          redact() {
+            sed -E \
+              -e 's/(RUSTFS_(ACCESS_KEY|SECRET_KEY)[=: ]+)[^[:space:]]+/\1[REDACTED]/Ig' \
+              -e 's/(Authorization:).*/\1 [REDACTED]/Ig' \
+              -e 's/(X-Amz-Signature=)[^&[:space:]]+/\1[REDACTED]/Ig' \
+              -e 's/^.*(password|secret|token)[=: ].*/[REDACTED SENSITIVE LINE]/Ig'
+          }
+          BODY_FILE="$(mktemp)"
+          {
+            echo "The **${SUITE_LABEL}** functional suite failed."
+            echo ""
+            echo "- Suite: \`${SUITE}\`"
+            echo "- Run: ${RUN_URL}"
+            echo "- Trigger: ${GITHUB_EVENT_NAME}"
+            echo "- Date: $(date -u +%Y-%m-%d)"
+            echo ""
+            echo "## Report (errors and symptoms)"
+            echo ""
+            if [ -s "${REPORT_FILE}" ]; then
+              redact < "${REPORT_FILE}"
+            elif [ -s "${LOG_FILE:-}" ]; then
+              echo "(report file missing; log tail below)"
+              echo ""
+              tail -n 200 "${LOG_FILE}" | redact
+            else
+              echo "(no report or log file was produced)"
+            fi
+          } | head -c 55000 > "${BODY_FILE}"
+          gh label create functional-test -R rustfs/backlog --color d73a4a 2>/dev/null || true
+          if ! gh issue create -R rustfs/backlog --title "${TITLE}" \
+              --body-file "${BODY_FILE}" --label functional-test; then
+            gh issue create -R rustfs/backlog --title "${TITLE}" --body-file "${BODY_FILE}"
+          fi
+          echo "filed backlog issue for suite ${SUITE}"
+
+      - name: Upload report and logs
+        if: always()
+        uses: actions/upload-artifact@b7c566a772e6b6bfb58ed0dc250532a479d7789f # v6
+        with:
+          name: rustfs-kms-test-${{ github.run_id }}
+          path: |
+            /tmp/rustfs-kms.log
+            /tmp/rustfs-kms-report.md
+          if-no-files-found: warn
+
+      - name: Cleanup environment (after)
+        if: always()
+        run: |
+          set -euo pipefail
+          read -r -a NODES <<< "${RUSTFS_NODES:-vm000 vm001 vm002}"
+          SSH_USER="${RUSTFS_SSH_USER:-azureuser}"
+          for node in "${NODES[@]}"; do
+            ssh -o BatchMode=yes -o ConnectTimeout=10 -o StrictHostKeyChecking=accept-new "${SSH_USER}@${node}" '
+              set -euo pipefail
+              SUDO=""; [ "$(id -u)" -ne 0 ] && SUDO="sudo -n"
+              ${SUDO} systemctl stop rustfs 2>/dev/null || true
+              if ${SUDO} dpkg -l rustfs 2>/dev/null | grep -q "^ii"; then
+                ${SUDO} dpkg -P rustfs
+              fi
+              for i in 1 2 3 4; do ${SUDO} rm -rf /data/rustfs${i}/mnmd; done
+              ${SUDO} rm -rf /var/log/rustfs /var/lib/rustfs/kms /var/lib/rustfs/kms-backup
+            '
+          done
+
+      - name: "Continue functional chain (next: Tier)"
+        # Only chain-triggered runs forward to the next suite; standalone
+        # workflow_dispatch runs stop after their own cleanup. A failed
+        # handoff must never pass silently: it retries, then files an alert
+        # issue in rustfs/backlog so a stalled chain is visible.
+        if: ${{ always() && github.event_name == 'repository_dispatch' }}
+        continue-on-error: true
+        env:
+          GH_TOKEN: ${{ secrets.PF_TESTING_GH_TOKEN }}
+        run: |
+          set -uo pipefail
+          if [ -z "${GH_TOKEN:-}" ]; then
+            echo "PF_TESTING_GH_TOKEN is not configured; cannot dispatch the next suite" >&2
+            exit 1
+          fi
+          DISPATCHED=0
+          for attempt in 1 2 3; do
+            if gh api --method POST repos/rustfs/rustfs/dispatches \
+              -f event_type='rustfs-chain-tier' \
+              -F 'client_payload[from_suite]=kms'; then
+              echo "dispatched next suite Tier (attempt ${attempt})"
+              DISPATCHED=1
+              break
+            fi
+            echo "dispatch attempt ${attempt} failed; retrying in ${attempt}0s" >&2
+            sleep "${attempt}0"
+          done
+          if [ "${DISPATCHED:-0}" -ne 1 ]; then
+            echo "ERROR: functional chain stalled: could not dispatch Tier after 3 attempts" >&2
+            TITLE="[functional][chain] stalled after kms (run ${GITHUB_RUN_ID})"
+            BODY_FILE="$(mktemp)"
+            {
+              echo "The functional chain could not hand off from **kms** to **Tier** after 3 attempts."
+              echo ""
+              echo "- Failed suite job: ${GITHUB_SERVER_URL}/${GITHUB_REPOSITORY}/actions/runs/${GITHUB_RUN_ID}"
+              echo "- Expected next event: 'rustfs-chain-tier'"
+              echo "- Likely cause: PF_TESTING_GH_TOKEN lacks contents:write on rustfs/rustfs, or the GitHub API was unavailable."
+              echo "- Recovery: re-dispatch manually with"
+              FENCE="$(printf "\x60\x60\x60")"; echo "  ${FENCE}"
+              echo "  gh api --method POST repos/rustfs/rustfs/dispatches -f event_type='rustfs-chain-tier'"
+              FENCE="$(printf "\x60\x60\x60")"; echo "  ${FENCE}"
+            } > "${BODY_FILE}"
+            gh issue create -R rustfs/backlog --title "${TITLE}" \
+              --body-file "${BODY_FILE}" --label functional-test \
+              || gh issue create -R rustfs/backlog --title "${TITLE}" --body-file "${BODY_FILE}" \
+              || echo "could not file the stall alert issue either; check the token" >&2
+            exit 1
+          fi
+
+      - name: Notify on failure
+        if: failure()
+        run: |
+          echo "RustFS KMS suite failed"
+          echo "See the uploaded report and log artifacts for details."
@@ -0,0 +1,309 @@
+name: RustFS Performance Test
+
+on:
+  workflow_dispatch:
+    inputs:
+      package_url:
+        description: 'Direct .deb URL (nightly/R2). Defaults to the latest nightly deb.'
+        required: false
+        type: string
+      test_method:
+        description: 'Benchmark method(s) to run (manual runs only; "all" = GET+PUT+MIXED)'
+        type: choice
+        options:
+          - all
+          - get
+          - put
+          - mixed
+        default: 'all'
+      object_size:
+        description: 'Object size(s) to test (manual runs only; "all" = all 10 sizes)'
+        type: choice
+        options:
+          - all
+          - 1KiB
+          - 4KiB
+          - 16KiB
+          - 128KiB
+          - 1MiB
+          - 4MiB
+          - 8MiB
+          - 16MiB
+          - 32MiB
+          - 64MiB
+        default: 'all'
+      warp_duration:
+        description: 'warp duration per round (e.g. 5m, 30s)'
+        required: false
+        default: '5m'
+      warp_concurrency:
+        description: 'warp concurrency'
+        required: false
+        default: '64'
+      cleanup_before:
+        description: 'Reset the nodes before the test (DESTROYS existing data/config)'
+        type: boolean
+        default: true
+      cleanup_after:
+        description: 'Reset the nodes after the test (DESTROYS test data/config)'
+        type: boolean
+        default: true
+  repository_dispatch:
+    # Chain entry: dispatched by rustfs-functional-chain.yml (runs on its own
+    # pf-testing runner, in parallel with the shared-VM chain).
+    types: [rustfs-chain-performance]
+
+permissions:
+  contents: read
+
+# Dedicated pf-testing runner/environment: own concurrency group so perf runs
+# never block (or are blocked by) the pool-expansion / heal tests.
+concurrency:
+  group: rustfs-performance-test
+  cancel-in-progress: false
+
+defaults:
+  run:
+    shell: bash
+
+env:
+  RUSTFS_ACCESS_KEY: ${{ secrets.RUSTFS_ACCESS_KEY }}
+  RUSTFS_SECRET_KEY: ${{ secrets.RUSTFS_SECRET_KEY }}
+  # Performance test uses its own node list (4 nodes); the shared
+  # RUSTFS_NODES secret is used by the 3-node pool-expansion / heal tests.
+  RUSTFS_NODES: ${{ secrets.RUSTFS_PERF_NODES || vars.RUSTFS_PERF_NODES || 'vm000 vm001 vm002 vm003' }}
+  RUSTFS_SSH_USER: ${{ secrets.RUSTFS_SSH_USER || vars.RUSTFS_SSH_USER }}
+  # Package used by the nightly run (workflow_dispatch inputs are empty for
+  # workflow_run events), i.e. the latest nightly deb published by nightly-gnu.yml.
+  RUSTFS_NIGHTLY_PACKAGE_URL: ${{ vars.RUSTFS_NIGHTLY_PACKAGE_URL || 'https://dl.rustfs.com/artifacts/rustfs/packages/nightly/rustfs-nightly-latest.deb' }}
+  # Fixed benchmark result directory so later steps can read summary.md
+  RUSTFS_RESULT_DIR: /tmp/rustfs-perf-results
+  # Cross-repo token for uploading reports to rustfs/dashboard (set in repo settings)
+  PF_TESTING_GH_TOKEN: ${{ secrets.PF_TESTING_GH_TOKEN }}
+
+jobs:
+  performance-test:
+    runs-on: pf-testing
+    # Requirement: a failing benchmark must not fail the workflow;
+    # failures are filed to rustfs/backlog.
+    continue-on-error: true
+    timeout-minutes: 900
+    # Run on manual dispatch, or when the nightly build completed successfully.
+    # Skipped when nightly failed.
+    if: ${{ github.event_name == 'workflow_dispatch' || github.event_name == 'repository_dispatch' }}
+    steps:
+      # auto-testing is private: clone it with the dedicated PF token (not
+      # GITHUB_TOKEN) and retry transient GitHub/network failures.
+      - name: Checkout auto-testing scripts (with retry)
+        env:
+          GH_TOKEN: ${{ secrets.PF_TESTING_GH_TOKEN }}
+        run: |
+          set -euo pipefail
+          rm -rf auto-testing
+          for attempt in 1 2 3 4 5; do
+            if gh repo clone rustfs/auto-testing auto-testing -- --depth 1 --quiet; then
+              echo "auto-testing cloned (attempt ${attempt})"
+              exit 0
+            fi
+            rm -rf auto-testing
+            echo "clone attempt ${attempt} failed; retrying in $((attempt * 15))s" >&2
+            sleep $((attempt * 15))
+          done
+          echo "ERROR: unable to clone rustfs/auto-testing after 5 attempts" >&2
+          exit 1
+
+      - name: Show environment
+        run: |
+          uname -a
+          jq --version
+          warp --version || true
+          df -h /data | tail -1
+
+      - name: Reset test environment (before)
+        if: ${{ inputs.cleanup_before != 'false' }}
+        run: |
+          chmod +x auto-testing/rustfs_performance_test.sh
+          ./auto-testing/rustfs_performance_test.sh --step 1 -y
+
+      - name: Install RustFS package & start cluster (4x4)
+        run: |
+          ARGS=(--steps "2,3,4" -y)
+          if [ -n "${{ inputs.package_url }}" ]; then
+            ARGS+=(--package-url "${{ inputs.package_url }}")
+          else
+            ARGS+=(--package-url "${{ env.RUSTFS_NIGHTLY_PACKAGE_URL }}")
+          fi
+          ./auto-testing/rustfs_performance_test.sh "${ARGS[@]}"
+
+      - name: Preflight checks
+        run: |
+          ARGS=(--preflight)
+          if [ -n "${{ inputs.package_url }}" ]; then
+            ARGS+=(--package-url "${{ inputs.package_url }}")
+          else
+            ARGS+=(--package-url "${{ env.RUSTFS_NIGHTLY_PACKAGE_URL }}")
+          fi
+          ./auto-testing/rustfs_performance_test.sh "${ARGS[@]}"
+
+      - name: Run benchmark (GET/PUT/MIXED)
+        id: benchmark
+        run: |
+          # Empty on automatic (workflow_run) runs -> full 30 rounds.
+          # Manual dispatch can restrict method(s)/size(s).
+          export WARP_METHODS="${{ inputs.test_method }}"
+          export WARP_SIZES="${{ inputs.object_size }}"
+          ./auto-testing/rustfs_performance_test.sh \
+            --step 5 -y \
+            --warp-duration "${{ inputs.warp_duration || '5m' }}" \
+            --warp-concurrency "${{ inputs.warp_concurrency || '64' }}" \
+            --log-file /tmp/rustfs-perf-test.log
+
+      - name: Analyze results
+        if: ${{ steps.benchmark.conclusion == 'success' }}
+        run: |
+          ./auto-testing/rustfs_performance_test.sh --step 6 -y
+
+      - name: Collect RustFS version info
+        if: ${{ steps.benchmark.conclusion == 'success' }}
+        env:
+          VERSION_FILE: /tmp/rustfs-version.txt
+        run: |
+          set -euo pipefail
+          read -r -a NODES <<< "${RUSTFS_NODES}"
+          [ "${#NODES[@]}" -gt 0 ] || { echo "RUSTFS_NODES is empty"; exit 1; }
+          NODE="${NODES[0]}"
+          SSH_USER="${RUSTFS_SSH_USER:-azureuser}"
+          {
+            echo "Node: ${NODE}"
+            echo "Command: rustfs --version"
+            echo ""
+            ssh -o BatchMode=yes -o ConnectTimeout=10 -o StrictHostKeyChecking=accept-new \
+              "${SSH_USER}@${NODE}" 'rustfs --version'
+          } > "${VERSION_FILE}"
+
+      - name: Upload report to dashboard (reports/YYYY-MM-DD.md)
+        if: ${{ steps.benchmark.conclusion == 'success' }}
+        env:
+          GH_TOKEN: ${{ env.PF_TESTING_GH_TOKEN }}
+          RESULT_DIR: ${{ env.RUSTFS_RESULT_DIR }}
+          VERSION_FILE: /tmp/rustfs-version.txt
+        run: |
+          set -euo pipefail
+          if [ -z "${GH_TOKEN:-}" ]; then
+            echo "PF_TESTING_GH_TOKEN is not configured; skipping report upload"
+            exit 0
+          fi
+          SUMMARY="${RESULT_DIR}/summary.md"
+          [ -f "${SUMMARY}" ] || { echo "summary.md not found at ${SUMMARY}"; exit 1; }
+          DATE="$(date -u +%Y-%m-%d)"
+          REPORT_PATH="reports/${DATE}.md"
+          {
+            echo "# RustFS nightly build performance testing report"
+            echo ""
+            echo "- **Date**: ${DATE}"
+            echo "- **Run**: ${{ github.server_url }}/${{ github.repository }}/actions/runs/${{ github.run_id }}"
+            echo "- **Trigger**: ${{ github.event_name }}"
+            echo "- **Package**: ${{ inputs.package_url || 'nightly (R2 latest)' }}"
+            echo ""
+            cat "${SUMMARY}"
+            echo ""
+            echo "## RustFS version"
+            echo '```text'
+            cat "${VERSION_FILE}"
+            echo '```'
+          } > /tmp/rustfs-perf-report.md
+          CONTENT="$(python3 -c 'import base64; print(base64.b64encode(open("/tmp/rustfs-perf-report.md","rb").read()).decode())')"
+          SHA="$(gh api "repos/rustfs/dashboard/contents/${REPORT_PATH}" -q '.sha' 2>/dev/null || true)"
+          if [ -n "${SHA}" ]; then
+            jq -n --arg msg "report: ${DATE}" --arg content "${CONTENT}" --arg sha "${SHA}" \
+              '{message:$msg, content:$content, sha:$sha}' \
+              | gh api --method PUT "repos/rustfs/dashboard/contents/${REPORT_PATH}" --input - >/dev/null
+            echo "updated ${REPORT_PATH} in rustfs/dashboard"
+          else
+            jq -n --arg msg "report: ${DATE}" --arg content "${CONTENT}" \
+              '{message:$msg, content:$content}' \
+              | gh api --method PUT "repos/rustfs/dashboard/contents/${REPORT_PATH}" --input - >/dev/null
+            echo "created ${REPORT_PATH} in rustfs/dashboard"
+          fi
+
+      - name: File failure issue in rustfs/backlog
+        if: ${{ always() && (failure() || steps.benchmark.outcome == 'failure' || steps.benchmark.outcome == 'cancelled') }}
+        continue-on-error: true
+        env:
+          GH_TOKEN: ${{ secrets.PF_TESTING_GH_TOKEN }}
+          SUITE: 'performance'
+          SUITE_LABEL: 'Performance'
+          RUN_URL: ${{ github.server_url }}/${{ github.repository }}/actions/runs/${{ github.run_id }}
+          REPORT_FILE: '/tmp/rustfs-perf-report.md'
+          LOG_FILE: '/tmp/rustfs-perf-test.log'
+        run: |
+          set -euo pipefail
+          if [ -z "${GH_TOKEN:-}" ]; then
+            echo "PF_TESTING_GH_TOKEN is not configured; skipping backlog issue"
+            exit 0
+          fi
+          TITLE="[functional][${SUITE}] ${SUITE_LABEL} suite failed (run ${GITHUB_RUN_ID})"
+          EXISTING="$(gh issue list -R rustfs/backlog --state all \
+            --search "in:title \"run ${GITHUB_RUN_ID}\"" \
+            --json number --jq '.[].number' || true)"
+          if [ -n "${EXISTING}" ]; then
+            echo "backlog issue already exists for run ${GITHUB_RUN_ID}; skipping"
+            exit 0
+          fi
+          redact() {
+            sed -E \
+              -e 's/(RUSTFS_(ACCESS_KEY|SECRET_KEY)[=: ]+)[^[:space:]]+/\1[REDACTED]/Ig' \
+              -e 's/(Authorization:).*/\1 [REDACTED]/Ig' \
+              -e 's/(X-Amz-Signature=)[^&[:space:]]+/\1[REDACTED]/Ig' \
+              -e 's/^.*(password|secret|token)[=: ].*/[REDACTED SENSITIVE LINE]/Ig'
+          }
+          BODY_FILE="$(mktemp)"
+          {
+            echo "The **${SUITE_LABEL}** functional suite failed."
+            echo ""
+            echo "- Suite: \`${SUITE}\`"
+            echo "- Run: ${RUN_URL}"
+            echo "- Trigger: ${GITHUB_EVENT_NAME}"
+            echo "- Date: $(date -u +%Y-%m-%d)"
+            echo ""
+            echo "## Report (errors and symptoms)"
+            echo ""
+            if [ -s "${REPORT_FILE}" ]; then
+              redact < "${REPORT_FILE}"
+            elif [ -s "${LOG_FILE:-}" ]; then
+              echo "(report file missing; log tail below)"
+              echo ""
+              tail -n 200 "${LOG_FILE}" | redact
+            else
+              echo "(no report or log file was produced)"
+            fi
+          } | head -c 55000 > "${BODY_FILE}"
+          gh label create functional-test -R rustfs/backlog --color d73a4a 2>/dev/null || true
+          if ! gh issue create -R rustfs/backlog --title "${TITLE}" \
+              --body-file "${BODY_FILE}" --label functional-test; then
+            gh issue create -R rustfs/backlog --title "${TITLE}" --body-file "${BODY_FILE}"
+          fi
+          echo "filed backlog issue for suite ${SUITE}"
+
+      - name: Upload test logs & results
+        if: always()
+        uses: actions/upload-artifact@b7c566a772e6b6bfb58ed0dc250532a479d7789f # v6
+        with:
+          name: rustfs-perf-test-${{ github.run_id }}
+          path: |
+            /tmp/rustfs-perf-test*.log
+            /tmp/rustfs-perf-results/**
+            /tmp/rustfs-version.txt
+          if-no-files-found: warn
+
+      - name: Reset test environment (after)
+        if: ${{ always() && inputs.cleanup_after != 'false' }}
+        run: |
+          ./auto-testing/rustfs_performance_test.sh --step 7 -y
+
+      - name: Notify on failure
+        if: failure()
+        run: |
+          echo "RustFS performance test failed"
+          echo "Package source: ${{ inputs.package_url || 'nightly (R2 latest)' }}"
+          echo "See the uploaded log artifact for details."
@@ -0,0 +1,651 @@
+name: RustFS Pool Expansion Test
+
+on:
+  workflow_dispatch:
+    inputs:
+      rustfs_version:
+        description: 'RustFS release tag to test (leave empty to use the latest nightly deb)'
+        required: false
+      package_url:
+        description: 'Direct .deb URL (nightly/R2/dev). Overrides rustfs_version.'
+        required: false
+        type: string
+      pools:
+        description: 'Number of pools to expand to (2 = first rebalance only)'
+        type: choice
+        options:
+          - '2'
+          - '3'
+        default: '3'
+      storage_threshold:
+        description: 'Stop writing when storage usage reaches N%'
+        required: false
+        default: '50'
+      warp_duration:
+        description: 'warp write duration (e.g. 5m, 10m)'
+        required: false
+        default: '10m'
+      warp_concurrent:
+        description: 'Pool fill: concurrent warp operations'
+        required: false
+        default: '32'
+      run_decommission:
+        description: 'Run the pool decommission step (3-pool topology only)'
+        type: boolean
+        default: true
+      cleanup_before:
+        description: 'Reset the nodes before the test (DESTROYS existing data/config)'
+        type: boolean
+        default: true
+      cleanup_after:
+        description: 'Reset the nodes after the test (DESTROYS test data/config)'
+        type: boolean
+        default: true
+  repository_dispatch:
+    # Chain handoff: dispatched when the heal suite finishes.
+    types: [rustfs-chain-pool]
+
+permissions:
+  contents: read
+
+# Only one test run at a time: the job mutates the same shared test
+# environment (vm000/vm001/vm002), so concurrent runs must not clobber each
+# other.
+concurrency:
+  group: rustfs-shared-functional-tests
+  cancel-in-progress: false
+
+defaults:
+  run:
+    shell: bash
+
+env:
+  RUSTFS_ACCESS_KEY: ${{ secrets.RUSTFS_ACCESS_KEY }}
+  RUSTFS_SECRET_KEY: ${{ secrets.RUSTFS_SECRET_KEY }}
+  RUSTFS_API_ENDPOINT: ${{ secrets.RUSTFS_API_ENDPOINT || vars.RUSTFS_API_ENDPOINT || vars.RUSTFS_RC_ENDPOINT }}
+  RUSTFS_NODES: ${{ secrets.RUSTFS_NODES || vars.RUSTFS_NODES }}
+  RUSTFS_SSH_USER: ${{ secrets.RUSTFS_SSH_USER || vars.RUSTFS_SSH_USER }}
+  PF_TESTING_GH_TOKEN: ${{ secrets.PF_TESTING_GH_TOKEN }}
+  # Package used by the nightly run (workflow_dispatch inputs are empty for
+  # workflow_run events), i.e. the latest nightly deb published by nightly-gnu.yml.
+  RUSTFS_NIGHTLY_PACKAGE_URL: ${{ vars.RUSTFS_NIGHTLY_PACKAGE_URL || 'https://dl.rustfs.com/artifacts/rustfs/packages/nightly/rustfs-nightly-latest.deb' }}
+
+jobs:
+  # Pool expansion: dispatched by the heal suite's chain handoff. Heal
+  # itself lives in rustfs-heal-test.yml and runs exactly once per chain.
+  pool-expansion-test:
+    name: Pool expansion / decommission test
+    runs-on: smoke-testing
+    # Requirement: a failing suite must not fail the workflow; failures
+    # are filed to rustfs/backlog and the chain continues.
+    continue-on-error: true
+    timeout-minutes: 360
+    if: ${{ github.event_name == 'workflow_dispatch' || github.event_name == 'repository_dispatch' }}
+    env:
+      RUSTFS_POOL_ADMIN_ENDPOINT: ${{ secrets.RUSTFS_POOL_ADMIN_ENDPOINT || vars.RUSTFS_POOL_ADMIN_ENDPOINT || 'http://rustfs-node1:9000' }}
+      RUSTFS_POOL_PROXY_ENDPOINT: http://127.0.0.1:19000
+      RUSTFS_POOL_WARP_ENDPOINT: http://127.0.0.1:19000
+      RUSTFS_SHARED_PROXY_ENDPOINT: ${{ secrets.RUSTFS_API_ENDPOINT || vars.RUSTFS_API_ENDPOINT || vars.RUSTFS_RC_ENDPOINT }}
+      RUSTFS_POOL_NODE_ENDPOINTS: ${{ secrets.RUSTFS_POOL_NODE_ENDPOINTS || vars.RUSTFS_POOL_NODE_ENDPOINTS || 'http://rustfs-node1:9000 http://rustfs-node2:9000 http://rustfs-node3:9000' }}
+    steps:
+      # auto-testing is private: clone it with the dedicated PF token (not
+      # GITHUB_TOKEN) and retry transient GitHub/network failures.
+      - name: Checkout auto-testing scripts (with retry)
+        env:
+          GH_TOKEN: ${{ secrets.PF_TESTING_GH_TOKEN }}
+        run: |
+          set -euo pipefail
+          rm -rf auto-testing
+          for attempt in 1 2 3 4 5; do
+            if gh repo clone rustfs/auto-testing auto-testing -- --depth 1 --quiet; then
+              echo "auto-testing cloned (attempt ${attempt})"
+              exit 0
+            fi
+            rm -rf auto-testing
+            echo "clone attempt ${attempt} failed; retrying in $((attempt * 15))s" >&2
+            sleep $((attempt * 15))
+          done
+          echo "ERROR: unable to clone rustfs/auto-testing after 5 attempts" >&2
+          exit 1
+
+      - name: Initialize pool test artifacts
+        run: |
+          set -euo pipefail
+          ARTIFACT_DIR="${RUNNER_TEMP}/rustfs-pool-${GITHUB_RUN_ID}-${GITHUB_RUN_ATTEMPT}"
+          mkdir -p "${ARTIFACT_DIR}"
+          echo "POOL_ARTIFACT_DIR=${ARTIFACT_DIR}" >> "${GITHUB_ENV}"
+
+      - name: Show environment
+        run: |
+          uname -a
+          jq --version
+          openssl version
+          warp --version || true
+          df -h /data | tail -1
+
+      - name: Cleanup environment (before)
+        if: ${{ inputs.cleanup_before != 'false' }}
+        run: |
+          set -euo pipefail
+          read -r -a NODES <<< "${RUSTFS_NODES:-vm000 vm001 vm002}"
+          SSH_USER="${RUSTFS_SSH_USER:-azureuser}"
+          for node in "${NODES[@]}"; do
+            ssh -o BatchMode=yes -o ConnectTimeout=10 -o StrictHostKeyChecking=accept-new "${SSH_USER}@${node}" '
+              set -euo pipefail
+              SUDO=""; [ "$(id -u)" -ne 0 ] && SUDO="sudo -n"
+              ${SUDO} systemctl stop rustfs 2>/dev/null || true
+              if ${SUDO} dpkg -l rustfs 2>/dev/null | grep -q "^ii"; then
+                ${SUDO} dpkg -P rustfs
+              fi
+              for i in 1 2 3 4; do ${SUDO} rm -rf /data/rustfs${i}/mnmd; done
+              ${SUDO} rm -rf /var/log/rustfs /var/lib/rustfs/kms /var/lib/rustfs/kms-backup
+            '
+          done
+
+      - name: Install RustFS package & start cluster
+        run: |
+          ARGS=(--steps "1,2,3" -y \
+            --admin-endpoint "${RUSTFS_POOL_ADMIN_ENDPOINT}" \
+            --warp-endpoint "${RUSTFS_POOL_WARP_ENDPOINT}" \
+            --node-endpoints "${RUSTFS_POOL_NODE_ENDPOINTS}" \
+            --log-file "${POOL_ARTIFACT_DIR}/pool-test.log")
+          if [ -n "${{ inputs.package_url }}" ]; then
+            ARGS+=(--package-url "${{ inputs.package_url }}")
+          elif [ -n "${{ inputs.rustfs_version }}" ]; then
+            ARGS+=(--version "${{ inputs.rustfs_version }}")
+          else
+            ARGS+=(--package-url "${{ env.RUSTFS_NIGHTLY_PACKAGE_URL }}")
+          fi
+          ./auto-testing/rustfs_pool_expand.sh "${ARGS[@]}"
+
+      - name: Preflight checks
+        run: |
+          ARGS=(--preflight \
+            --admin-endpoint "${RUSTFS_POOL_ADMIN_ENDPOINT}" \
+            --warp-endpoint "${RUSTFS_POOL_WARP_ENDPOINT}" \
+            --node-endpoints "${RUSTFS_POOL_NODE_ENDPOINTS}" \
+            --log-file "${POOL_ARTIFACT_DIR}/pool-test.log")
+          if [ -n "${{ inputs.package_url }}" ]; then
+            ARGS+=(--package-url "${{ inputs.package_url }}")
+          elif [ -n "${{ inputs.rustfs_version }}" ]; then
+            ARGS+=(--version "${{ inputs.rustfs_version }}")
+          else
+            ARGS+=(--package-url "${{ env.RUSTFS_NIGHTLY_PACKAGE_URL }}")
+          fi
+          ./auto-testing/rustfs_pool_expand.sh "${ARGS[@]}"
+
+      - name: Reset dedicated pool proxy
+        run: |
+          set -euo pipefail
+          RUSTFS_POOL_NGINX_CONFIG_PATH=/etc/nginx/conf.d/rustfs-pool-test.conf \
+          RUSTFS_POOL_NGINX_LISTEN="${RUSTFS_POOL_PROXY_ENDPOINT#http://}" \
+          RUSTFS_POOL_NGINX_ACCESS_LOG=/var/log/nginx/rustfs-pool-test-access.log \
+          RUSTFS_POOL_NGINX_ERROR_LOG=/var/log/nginx/rustfs-pool-test-error.log \
+            ./auto-testing/rustfs_pool_nginx_stage.sh cleanup
+
+      - name: Capture pool test baseline
+        run: |
+          set -uo pipefail
+          BASELINE_FILE="${POOL_ARTIFACT_DIR}/pool-baseline.log"
+          read -r -a NODES <<< "${RUSTFS_NODES:-vm000 vm001 vm002}"
+          read -r -a DIRECT_ENDPOINTS <<< "${RUSTFS_POOL_NODE_ENDPOINTS}"
+          SSH_USER="${RUSTFS_SSH_USER:-azureuser}"
+          failed=0
+          : > "${BASELINE_FILE}"
+
+          if [ "${#DIRECT_ENDPOINTS[@]}" -lt "${#NODES[@]}" ]; then
+            echo "not enough direct endpoints for the configured nodes" | tee -a "${BASELINE_FILE}" >&2
+            exit 1
+          fi
+
+          for index in "${!NODES[@]}"; do
+            node="${NODES[$index]}"
+            endpoint="${DIRECT_ENDPOINTS[$index]}"
+            body_file="${POOL_ARTIFACT_DIR}/ready-baseline-$((index + 1)).body"
+            {
+              echo "--- node=${node} endpoint=${endpoint} ---"
+              if ! ssh -o BatchMode=yes -o ConnectTimeout=10 -o StrictHostKeyChecking=accept-new \
+                "${SSH_USER}@${node}" '
+                  set -euo pipefail
+                  SUDO=""; [ "$(id -u)" -ne 0 ] && SUDO="sudo -n"
+                  echo "--- rustfs version ---"
+                  rustfs --version
+                  echo "--- systemd state ---"
+                  ${SUDO} systemctl show rustfs --no-pager \
+                    --property=ActiveState,SubState,Result,ExecMainPID,ExecMainStartTimestamp,NRestarts
+                '; then
+                echo "baseline collection failed for ${node}"
+                failed=1
+              fi
+              curl -sS --connect-timeout 5 --max-time 15 -o "${body_file}" \
+                -w "baseline_ready=${endpoint} http=%{http_code} connect=%{time_connect} ttfb=%{time_starttransfer} total=%{time_total}\n" \
+                "${endpoint%/}/health/ready" || true
+              echo "--- readiness body ---"
+              cat "${body_file}" 2>/dev/null || true
+              echo
+            } >> "${BASELINE_FILE}" 2>&1
+          done
+
+          [ "${failed}" -eq 0 ] || exit 1
+
+      - name: Run pool expansion & decommission test
+        id: pool_test
+        run: |
+          set -o pipefail
+          STEPS="4,5,6"
+          if [ "${{ inputs.pools || '3' }}" = "3" ]; then
+            STEPS="$STEPS,7,8"
+            if [ "${{ inputs.run_decommission != 'false' }}" = "true" ]; then
+              STEPS="$STEPS,9"
+            fi
+          fi
+          ARGS=(--steps "$STEPS" --with-warp -y \
+            --admin-endpoint "${RUSTFS_POOL_ADMIN_ENDPOINT}" \
+            --warp-endpoint "${RUSTFS_POOL_WARP_ENDPOINT}" \
+            --node-endpoints "${RUSTFS_POOL_NODE_ENDPOINTS}" \
+            --storage-threshold "${{ inputs.storage_threshold || '50' }}" \
+            --warp-duration "${{ inputs.warp_duration || '10m' }}" \
+            --warp-concurrent "${{ inputs.warp_concurrent || '32' }}" \
+            --log-file "${POOL_ARTIFACT_DIR}/pool-test.log")
+          if [ -n "${RUSTFS_POOL_PROXY_ENDPOINT}" ]; then
+            ARGS+=(--proxy-endpoint "${RUSTFS_POOL_PROXY_ENDPOINT}")
+          fi
+          if [ -n "${{ inputs.package_url }}" ]; then
+            ARGS+=(--package-url "${{ inputs.package_url }}")
+          elif [ -n "${{ inputs.rustfs_version }}" ]; then
+            ARGS+=(--version "${{ inputs.rustfs_version }}")
+          else
+            ARGS+=(--package-url "${{ env.RUSTFS_NIGHTLY_PACKAGE_URL }}")
+          fi
+          RUSTFS_WARP_LOG_FILE="${POOL_ARTIFACT_DIR}/warp.log" \
+          RUSTFS_PROXY_STAGE_HOOK=./auto-testing/rustfs_pool_nginx_stage.sh \
+          RUSTFS_POOL_NGINX_CONFIG_PATH=/etc/nginx/conf.d/rustfs-pool-test.conf \
+          RUSTFS_POOL_NGINX_LISTEN="${RUSTFS_POOL_PROXY_ENDPOINT#http://}" \
+          RUSTFS_POOL_NGINX_ACCESS_LOG=/var/log/nginx/rustfs-pool-test-access.log \
+          RUSTFS_POOL_NGINX_ERROR_LOG=/var/log/nginx/rustfs-pool-test-error.log \
+            ./auto-testing/rustfs_pool_expand.sh "${ARGS[@]}"
+
+      - name: Collect pool test diagnostics
+        if: always()
+        run: |
+          set -uo pipefail
+          ARTIFACT_DIR="${POOL_ARTIFACT_DIR:-${RUNNER_TEMP}/rustfs-pool-${GITHUB_RUN_ID}-${GITHUB_RUN_ATTEMPT}}"
+          mkdir -p "${ARTIFACT_DIR}"
+          echo "POOL_ARTIFACT_DIR=${ARTIFACT_DIR}" >> "${GITHUB_ENV}"
+
+          redact() {
+            sed -E \
+              -e 's/(RUSTFS_(ACCESS_KEY|SECRET_KEY)=).*/\1[REDACTED]/Ig' \
+              -e 's/(Authorization:).*/\1 [REDACTED]/Ig' \
+              -e 's/(X-Amz-Signature=)[^&[:space:]]+/\1[REDACTED]/Ig' \
+              -e 's/(proxy_set_header[[:space:]]+Authorization[[:space:]]+).*/\1[REDACTED];/Ig' \
+              -e 's/^.*(password|secret|token).*/[REDACTED SENSITIVE LINE]/Ig'
+          }
+
+          if [ "$(id -u)" -eq 0 ]; then
+            SUDO=()
+          else
+            SUDO=(sudo -n)
+          fi
+
+          {
+            echo "captured_at=$(date -u +%Y-%m-%dT%H:%M:%SZ)"
+            echo "run_id=${GITHUB_RUN_ID}"
+            echo "run_attempt=${GITHUB_RUN_ATTEMPT}"
+            if command -v nginx >/dev/null 2>&1; then
+              "${SUDO[@]}" nginx -T 2>&1 || echo "nginx -T failed"
+            else
+              echo "nginx is not installed on the runner"
+            fi
+          } | redact > "${ARTIFACT_DIR}/nginx-config-redacted.txt"
+
+          for log_path in \
+            /var/log/nginx/access.log \
+            /var/log/nginx/error.log \
+            /var/log/nginx/rustfs-pool-test-access.log \
+            /var/log/nginx/rustfs-pool-test-error.log; do
+            log_name="$(basename "${log_path}")"
+            if "${SUDO[@]}" test -r "${log_path}" 2>/dev/null; then
+              "${SUDO[@]}" cat "${log_path}" 2>&1 | redact \
+                > "${ARTIFACT_DIR}/nginx-${log_name%.log}-redacted.log"
+            else
+              echo "unavailable: ${log_path}" > "${ARTIFACT_DIR}/nginx-${log_name%.log}-redacted.log"
+            fi
+          done
+          "${SUDO[@]}" journalctl -u nginx --no-pager -n 5000 2>&1 | redact \
+            > "${ARTIFACT_DIR}/nginx-journal-redacted.log" || true
+
+          read -r -a NODES <<< "${RUSTFS_NODES:-vm000 vm001 vm002}"
+          SSH_USER="${RUSTFS_SSH_USER:-azureuser}"
+          for node in "${NODES[@]}"; do
+            safe_node="${node//[^A-Za-z0-9_.-]/_}"
+            {
+              if ! ssh -o BatchMode=yes -o ConnectTimeout=10 -o StrictHostKeyChecking=accept-new \
+                "${SSH_USER}@${node}" '
+                  SUDO=""; [ "$(id -u)" -ne 0 ] && SUDO="sudo -n"
+                  echo "--- rustfs version ---"
+                  rustfs --version 2>&1 || true
+                  echo "--- systemd state ---"
+                  ${SUDO} systemctl show rustfs --no-pager \
+                    --property=ActiveState,SubState,Result,ExecMainPID,ExecMainStartTimestamp,NRestarts 2>&1 || true
+                  echo "--- rustfs journal ---"
+                  ${SUDO} journalctl -u rustfs --no-pager -n 10000 2>&1 || true
+                  echo "--- rustfs file logs ---"
+                  if ${SUDO} test -d /var/log/rustfs; then
+                    ${SUDO} find /var/log/rustfs -maxdepth 2 -type f -print 2>/dev/null | while IFS= read -r file; do
+                      echo "--- ${file} (last 5000 lines) ---"
+                      ${SUDO} tail -n 5000 "${file}" 2>&1 || true
+                    done
+                  else
+                    echo "/var/log/rustfs is unavailable"
+                  fi
+                '; then
+                echo "SSH diagnostics failed for ${node}"
+              fi
+            } 2>&1 | redact > "${ARTIFACT_DIR}/${safe_node}-rustfs-redacted.log"
+          done
+
+          : > "${ARTIFACT_DIR}/endpoint-ready-probes.log"
+          read -r -a DIRECT_ENDPOINTS <<< "${RUSTFS_POOL_NODE_ENDPOINTS}"
+          probe_index=0
+          for endpoint in "${DIRECT_ENDPOINTS[@]}"; do
+            probe_index=$((probe_index + 1))
+            curl -sS --connect-timeout 5 --max-time 15 -o "${ARTIFACT_DIR}/ready-direct-${probe_index}.body" \
+              -w "direct[${probe_index}]=${endpoint} http=%{http_code} connect=%{time_connect} ttfb=%{time_starttransfer} total=%{time_total}\n" \
+              "${endpoint%/}/health/ready" >> "${ARTIFACT_DIR}/endpoint-ready-probes.log" 2>&1 || true
+          done
+          if [ -n "${RUSTFS_POOL_PROXY_ENDPOINT}" ]; then
+            curl -sS --connect-timeout 5 --max-time 15 -o "${ARTIFACT_DIR}/ready-proxy.body" \
+              -w "proxy=${RUSTFS_POOL_PROXY_ENDPOINT} http=%{http_code} connect=%{time_connect} ttfb=%{time_starttransfer} total=%{time_total}\n" \
+              "${RUSTFS_POOL_PROXY_ENDPOINT%/}/health/ready" >> "${ARTIFACT_DIR}/endpoint-ready-probes.log" 2>&1 || true
+          fi
+          if [ -n "${RUSTFS_SHARED_PROXY_ENDPOINT}" ]; then
+            curl -sS --connect-timeout 5 --max-time 15 -o "${ARTIFACT_DIR}/ready-shared-proxy.body" \
+              -w "shared_proxy=${RUSTFS_SHARED_PROXY_ENDPOINT} http=%{http_code} connect=%{time_connect} ttfb=%{time_starttransfer} total=%{time_total}\n" \
+              "${RUSTFS_SHARED_PROXY_ENDPOINT%/}/health/ready" >> "${ARTIFACT_DIR}/endpoint-ready-probes.log" 2>&1 || true
+          fi
+
+      - name: Generate report
+        if: always()
+        run: |
+          set -euo pipefail
+          LOG_FILE="${POOL_ARTIFACT_DIR}/pool-test.log"
+          REPORT_FILE="${POOL_ARTIFACT_DIR}/pool-report.md"
+          PACKAGE_URL='${{ inputs.package_url }}'
+          RUSTFS_VERSION='${{ inputs.rustfs_version }}'
+          if [ -n "${PACKAGE_URL}" ]; then
+            PACKAGE_SOURCE="${PACKAGE_URL}"
+          elif [ -n "${RUSTFS_VERSION}" ]; then
+            PACKAGE_SOURCE="version ${RUSTFS_VERSION}"
+          else
+            PACKAGE_SOURCE="${RUSTFS_NIGHTLY_PACKAGE_URL}"
+          fi
+          {
+            echo "# RustFS pool expansion test report"
+            echo ""
+            echo "- Run: ${{ github.server_url }}/${{ github.repository }}/actions/runs/${{ github.run_id }}"
+            echo "- Trigger: ${{ github.event_name }}"
+            echo "- Package: ${PACKAGE_SOURCE}"
+            echo "- Warp concurrent: ${{ inputs.warp_concurrent || '32' }}"
+            echo "- Test Step Outcome: ${{ steps.pool_test.outcome }}"
+            echo ""
+            echo "## Log tail"
+            echo '```text'
+            tail -n 200 "${LOG_FILE}" || true
+            echo '```'
+          } | tee "${REPORT_FILE}"
+          cat "${REPORT_FILE}" >> "${GITHUB_STEP_SUMMARY}"
+
+      - name: Validate pool diagnostic completeness
+        if: always()
+        run: |
+          set -euo pipefail
+          failed=0
+          require_nonempty() {
+            if [ ! -s "$1" ]; then
+              echo "required diagnostic is missing or empty: $1" >&2
+              failed=1
+            fi
+          }
+          require_available() {
+            if [ ! -e "$1" ]; then
+              echo "required diagnostic is missing: $1" >&2
+              failed=1
+            elif grep -Fq 'unavailable:' "$1" 2>/dev/null; then
+              echo "required diagnostic could not be collected: $1" >&2
+              failed=1
+            fi
+          }
+
+          require_nonempty "${POOL_ARTIFACT_DIR}/pool-test.log"
+          require_nonempty "${POOL_ARTIFACT_DIR}/warp.log"
+          require_nonempty "${POOL_ARTIFACT_DIR}/pool-report.md"
+          require_nonempty "${POOL_ARTIFACT_DIR}/pool-baseline.log"
+          require_nonempty "${POOL_ARTIFACT_DIR}/nginx-config-redacted.txt"
+          require_nonempty "${POOL_ARTIFACT_DIR}/nginx-rustfs-pool-test-access-redacted.log"
+          require_available "${POOL_ARTIFACT_DIR}/nginx-rustfs-pool-test-access-redacted.log"
+          require_available "${POOL_ARTIFACT_DIR}/nginx-rustfs-pool-test-error-redacted.log"
+          require_nonempty "${POOL_ARTIFACT_DIR}/endpoint-ready-probes.log"
+
+          read -r -a NODES <<< "${RUSTFS_NODES:-vm000 vm001 vm002}"
+          if grep -Fq 'baseline collection failed' "${POOL_ARTIFACT_DIR}/pool-baseline.log" 2>/dev/null; then
+            echo "one or more node baselines could not be collected" >&2
+            failed=1
+          fi
+          for node in "${NODES[@]}"; do
+            safe_node="${node//[^A-Za-z0-9_.-]/_}"
+            node_log="${POOL_ARTIFACT_DIR}/${safe_node}-rustfs-redacted.log"
+            require_nonempty "${node_log}"
+            if grep -Fq "SSH diagnostics failed for ${node}" "${node_log}" 2>/dev/null; then
+              echo "node diagnostics failed: ${node_log}" >&2
+              failed=1
+            fi
+            if ! grep -Eq '^rustfs @' "${node_log}" 2>/dev/null \
+              || ! grep -Eq '^NRestarts=[0-9]+$' "${node_log}" 2>/dev/null; then
+              echo "node version or restart evidence is incomplete: ${node_log}" >&2
+              failed=1
+            elif grep -Eq '^NRestarts=[1-9][0-9]*$' "${node_log}"; then
+              echo "RustFS restarted unexpectedly during the run: ${node_log}" >&2
+              failed=1
+            fi
+          done
+
+          if ! grep -Fq "upstream_status=\"\$upstream_status\"" \
+            "${POOL_ARTIFACT_DIR}/nginx-config-redacted.txt"; then
+            echo "Nginx config does not expose upstream status fields" >&2
+            failed=1
+          fi
+          if ! grep -Eq '^proxy=.* http=200([[:space:]]|$)' "${POOL_ARTIFACT_DIR}/endpoint-ready-probes.log"; then
+            echo "dedicated proxy readiness probe did not return HTTP 200" >&2
+            failed=1
+          fi
+          if grep -Eq 'status=50(2|4)|upstream_status="[^"]*50(2|4)' \
+            "${POOL_ARTIFACT_DIR}/nginx-rustfs-pool-test-access-redacted.log"; then
+            echo "dedicated proxy access log contains a 502/504 response" >&2
+            failed=1
+          fi
+          if grep -Eiq 'upstream prematurely closed connection|upstream timed out|(connect\(\)|recv\(\)|send\(\)) failed.*upstream|connection reset by peer.*upstream' \
+            "${POOL_ARTIFACT_DIR}/nginx-rustfs-pool-test-error-redacted.log"; then
+            echo "dedicated proxy error log contains an upstream timeout or connection failure" >&2
+            failed=1
+          fi
+          [ "${failed}" -eq 0 ] || exit 1
+
+      - name: Upload functional report to dashboard
+        if: always()
+        continue-on-error: true
+        env:
+          GH_TOKEN: ${{ env.PF_TESTING_GH_TOKEN }}
+          SUITE: pool
+        run: |
+          set -euo pipefail
+          REPORT_FILE="${POOL_ARTIFACT_DIR}/pool-report.md"
+          if [ -z "${GH_TOKEN:-}" ]; then
+            echo "PF_TESTING_GH_TOKEN is not configured; skipping dashboard upload"
+            exit 0
+          fi
+          DATE="$(date -u +%Y-%m-%d)"
+          REPORT_PATH="functional-reports/${SUITE}/${DATE}.md"
+          CONTENT="$(python3 -c 'import base64,sys;print(base64.b64encode(open(sys.argv[1],"rb").read()).decode())' "${REPORT_FILE}")"
+          SHA="$(gh api "repos/rustfs/dashboard/contents/${REPORT_PATH}" -q '.sha' 2>/dev/null || true)"
+          if [ -n "${SHA}" ]; then
+            jq -n --arg msg "report(${SUITE}): ${DATE}" --arg content "${CONTENT}" --arg sha "${SHA}" \
+              '{message:$msg, content:$content, sha:$sha}' \
+              | gh api --method PUT "repos/rustfs/dashboard/contents/${REPORT_PATH}" --input - >/dev/null
+          else
+            jq -n --arg msg "report(${SUITE}): ${DATE}" --arg content "${CONTENT}" \
+              '{message:$msg, content:$content}' \
+              | gh api --method PUT "repos/rustfs/dashboard/contents/${REPORT_PATH}" --input - >/dev/null
+          fi
+
+      - name: File failure issue in rustfs/backlog
+        if: ${{ always() && (failure() || steps.pool_test.outcome == 'failure' || steps.pool_test.outcome == 'cancelled') }}
+        continue-on-error: true
+        env:
+          GH_TOKEN: ${{ secrets.PF_TESTING_GH_TOKEN }}
+          SUITE: 'pool'
+          SUITE_LABEL: 'Pool expansion'
+          RUN_URL: ${{ github.server_url }}/${{ github.repository }}/actions/runs/${{ github.run_id }}
+          REPORT_FILE: '${{ env.POOL_ARTIFACT_DIR }}/pool-report.md'
+          LOG_FILE: '${{ env.POOL_ARTIFACT_DIR }}/pool-test.log'
+        run: |
+          set -euo pipefail
+          if [ -z "${GH_TOKEN:-}" ]; then
+            echo "PF_TESTING_GH_TOKEN is not configured; skipping backlog issue"
+            exit 0
+          fi
+          TITLE="[functional][${SUITE}] ${SUITE_LABEL} suite failed (run ${GITHUB_RUN_ID})"
+          EXISTING="$(gh issue list -R rustfs/backlog --state all \
+            --search "in:title \"run ${GITHUB_RUN_ID}\"" \
+            --json number --jq '.[].number' || true)"
+          if [ -n "${EXISTING}" ]; then
+            echo "backlog issue already exists for run ${GITHUB_RUN_ID}; skipping"
+            exit 0
+          fi
+          redact() {
+            sed -E \
+              -e 's/(RUSTFS_(ACCESS_KEY|SECRET_KEY)[=: ]+)[^[:space:]]+/\1[REDACTED]/Ig' \
+              -e 's/(Authorization:).*/\1 [REDACTED]/Ig' \
+              -e 's/(X-Amz-Signature=)[^&[:space:]]+/\1[REDACTED]/Ig' \
+              -e 's/^.*(password|secret|token)[=: ].*/[REDACTED SENSITIVE LINE]/Ig'
+          }
+          BODY_FILE="$(mktemp)"
+          {
+            echo "The **${SUITE_LABEL}** functional suite failed."
+            echo ""
+            echo "- Suite: \`${SUITE}\`"
+            echo "- Run: ${RUN_URL}"
+            echo "- Trigger: ${GITHUB_EVENT_NAME}"
+            echo "- Date: $(date -u +%Y-%m-%d)"
+            echo ""
+            echo "## Report (errors and symptoms)"
+            echo ""
+            if [ -s "${REPORT_FILE}" ]; then
+              redact < "${REPORT_FILE}"
+            elif [ -s "${LOG_FILE:-}" ]; then
+              echo "(report file missing; log tail below)"
+              echo ""
+              tail -n 200 "${LOG_FILE}" | redact
+            else
+              echo "(no report or log file was produced)"
+            fi
+          } | head -c 55000 > "${BODY_FILE}"
+          gh label create functional-test -R rustfs/backlog --color d73a4a 2>/dev/null || true
+          if ! gh issue create -R rustfs/backlog --title "${TITLE}" \
+              --body-file "${BODY_FILE}" --label functional-test; then
+            gh issue create -R rustfs/backlog --title "${TITLE}" --body-file "${BODY_FILE}"
+          fi
+          echo "filed backlog issue for suite ${SUITE}"
+
+      - name: Upload test logs
+        if: always()
+        uses: actions/upload-artifact@b7c566a772e6b6bfb58ed0dc250532a479d7789f # v6
+        with:
+          name: rustfs-pool-test-${{ github.run_id }}-${{ github.run_attempt }}
+          path: ${{ runner.temp }}/rustfs-pool-${{ github.run_id }}-${{ github.run_attempt }}
+          if-no-files-found: warn
+
+      - name: Restore dedicated pool proxy
+        if: always()
+        run: |
+          set -euo pipefail
+          RUSTFS_POOL_NGINX_CONFIG_PATH=/etc/nginx/conf.d/rustfs-pool-test.conf \
+          RUSTFS_POOL_NGINX_LISTEN="${RUSTFS_POOL_PROXY_ENDPOINT#http://}" \
+          RUSTFS_POOL_NGINX_ACCESS_LOG=/var/log/nginx/rustfs-pool-test-access.log \
+          RUSTFS_POOL_NGINX_ERROR_LOG=/var/log/nginx/rustfs-pool-test-error.log \
+            ./auto-testing/rustfs_pool_nginx_stage.sh cleanup
+
+      - name: Cleanup environment (after)
+        if: ${{ always() && inputs.cleanup_after != 'false' }}
+        run: |
+          set -euo pipefail
+          read -r -a NODES <<< "${RUSTFS_NODES:-vm000 vm001 vm002}"
+          SSH_USER="${RUSTFS_SSH_USER:-azureuser}"
+          for node in "${NODES[@]}"; do
+            ssh -o BatchMode=yes -o ConnectTimeout=10 -o StrictHostKeyChecking=accept-new "${SSH_USER}@${node}" '
+              set -euo pipefail
+              SUDO=""; [ "$(id -u)" -ne 0 ] && SUDO="sudo -n"
+              ${SUDO} systemctl stop rustfs 2>/dev/null || true
+              if ${SUDO} dpkg -l rustfs 2>/dev/null | grep -q "^ii"; then
+                ${SUDO} dpkg -P rustfs
+              fi
+              for i in 1 2 3 4; do ${SUDO} rm -rf /data/rustfs${i}/mnmd; done
+              ${SUDO} rm -rf /var/log/rustfs /var/lib/rustfs/kms /var/lib/rustfs/kms-backup
+            '
+          done
+
+      - name: "Continue functional chain (next: Security)"
+        # Only chain-triggered runs forward to the next suite; standalone
+        # workflow_dispatch runs stop after their own cleanup. A failed
+        # handoff must never pass silently: it retries, then files an alert
+        # issue in rustfs/backlog so a stalled chain is visible.
+        if: ${{ always() && github.event_name == 'repository_dispatch' }}
+        continue-on-error: true
+        env:
+          GH_TOKEN: ${{ secrets.PF_TESTING_GH_TOKEN }}
+        run: |
+          set -uo pipefail
+          if [ -z "${GH_TOKEN:-}" ]; then
+            echo "PF_TESTING_GH_TOKEN is not configured; cannot dispatch the next suite" >&2
+            exit 1
+          fi
+          DISPATCHED=0
+          for attempt in 1 2 3; do
+            if gh api --method POST repos/rustfs/rustfs/dispatches \
+              -f event_type='rustfs-chain-security' \
+              -F 'client_payload[from_suite]=pool'; then
+              echo "dispatched next suite Security (attempt ${attempt})"
+              DISPATCHED=1
+              break
+            fi
+            echo "dispatch attempt ${attempt} failed; retrying in ${attempt}0s" >&2
+            sleep "${attempt}0"
+          done
+          if [ "${DISPATCHED:-0}" -ne 1 ]; then
+            echo "ERROR: functional chain stalled: could not dispatch Security after 3 attempts" >&2
+            TITLE="[functional][chain] stalled after pool (run ${GITHUB_RUN_ID})"
+            BODY_FILE="$(mktemp)"
+            {
+              echo "The functional chain could not hand off from **pool** to **Security** after 3 attempts."
+              echo ""
+              echo "- Failed suite job: ${GITHUB_SERVER_URL}/${GITHUB_REPOSITORY}/actions/runs/${GITHUB_RUN_ID}"
+              echo "- Expected next event: 'rustfs-chain-security'"
+              echo "- Likely cause: PF_TESTING_GH_TOKEN lacks contents:write on rustfs/rustfs, or the GitHub API was unavailable."
+              echo "- Recovery: re-dispatch manually with"
+              FENCE="$(printf "\x60\x60\x60")"; echo "  ${FENCE}"
+              echo "  gh api --method POST repos/rustfs/rustfs/dispatches -f event_type='rustfs-chain-security'"
+              FENCE="$(printf "\x60\x60\x60")"; echo "  ${FENCE}"
+            } > "${BODY_FILE}"
+            gh issue create -R rustfs/backlog --title "${TITLE}" \
+              --body-file "${BODY_FILE}" --label functional-test \
+              || gh issue create -R rustfs/backlog --title "${TITLE}" --body-file "${BODY_FILE}" \
+              || echo "could not file the stall alert issue either; check the token" >&2
+            exit 1
+          fi
+
+      - name: Notify on failure
+        if: failure()
+        run: |
+          echo "RustFS pool expansion test failed"
+          echo "Package source: ${{ inputs.package_url || inputs.rustfs_version || 'nightly (R2 latest)' }}"
+          echo "See the uploaded log artifact for details."
@@ -0,0 +1,365 @@
+# Copyright 2024 RustFS Team
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#     http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+
+name: RustFS Replication Test
+
+on:
+  workflow_dispatch:
+    inputs:
+      rustfs_version:
+        description: 'RustFS release tag to test (e.g. 1.0.0-rc.4-preview.1)'
+        required: false
+        default: '1.0.0-rc.4-preview.1'
+      package_url:
+        description: 'Direct .deb URL (nightly/R2/dev). Overrides rustfs_version.'
+        required: false
+        type: string
+      suite:
+        description: 'Suite to run (all = bucket REP-* then site SITE-*)'
+        type: choice
+        options:
+          - all
+          - bucket
+          - site
+        default: all
+  repository_dispatch:
+    # Chain handoff: dispatched when the security suite finishes. This is the
+    # last link of the functional chain.
+    types: [rustfs-chain-replication]
+
+permissions:
+  contents: read
+
+# The replication suite uses the same shared VMs as the other functional
+# tests, so it must serialize with them instead of running in parallel.
+concurrency:
+  group: rustfs-shared-functional-tests
+  cancel-in-progress: false
+
+defaults:
+  run:
+    shell: bash
+
+env:
+  RUSTFS_ACCESS_KEY: ${{ secrets.RUSTFS_ACCESS_KEY }}
+  RUSTFS_SECRET_KEY: ${{ secrets.RUSTFS_SECRET_KEY }}
+  RUSTFS_NODES: ${{ secrets.RUSTFS_NODES || vars.RUSTFS_NODES }}
+  RUSTFS_SSH_USER: ${{ secrets.RUSTFS_SSH_USER || vars.RUSTFS_SSH_USER }}
+  RUSTFS_NIGHTLY_PACKAGE_URL: ${{ vars.RUSTFS_NIGHTLY_PACKAGE_URL || 'https://dl.rustfs.com/artifacts/rustfs/packages/nightly/rustfs-nightly-latest.deb' }}
+  PF_TESTING_GH_TOKEN: ${{ secrets.PF_TESTING_GH_TOKEN }}
+
+jobs:
+  replication-test:
+    runs-on: smoke-testing
+    # A failed replication run must not break the chain or the workflow: the
+    # failure is reported to rustfs/backlog instead (see the issue step).
+    continue-on-error: true
+    timeout-minutes: 360
+    if: ${{ github.event_name == 'workflow_dispatch' || github.event_name == 'repository_dispatch' }}
+    steps:
+      # auto-testing is private: clone it with the dedicated PF token (not
+      # GITHUB_TOKEN) and retry transient GitHub/network failures.
+      - name: Checkout auto-testing scripts (with retry)
+        env:
+          GH_TOKEN: ${{ secrets.PF_TESTING_GH_TOKEN }}
+        run: |
+          set -euo pipefail
+          rm -rf auto-testing
+          for attempt in 1 2 3 4 5; do
+            if gh repo clone rustfs/auto-testing auto-testing -- --depth 1 --quiet; then
+              echo "auto-testing cloned (attempt ${attempt})"
+              exit 0
+            fi
+            rm -rf auto-testing
+            echo "clone attempt ${attempt} failed; retrying in $((attempt * 15))s" >&2
+            sleep $((attempt * 15))
+          done
+          echo "ERROR: unable to clone rustfs/auto-testing after 5 attempts" >&2
+          exit 1
+
+      - name: Show environment
+        run: |
+          uname -a
+          jq --version
+          openssl version
+          aws --version
+          df -h /data | tail -1 || true
+
+      - name: Cleanup environment (before)
+        run: |
+          set -euo pipefail
+          read -r -a NODES <<< "${RUSTFS_NODES:-vm000 vm001 vm002}"
+          SSH_USER="${RUSTFS_SSH_USER:-azureuser}"
+          for node in "${NODES[@]}"; do
+            ssh -o BatchMode=yes -o ConnectTimeout=10 -o StrictHostKeyChecking=accept-new "${SSH_USER}@${node}" '
+              set -euo pipefail
+              SUDO=""; [ "$(id -u)" -ne 0 ] && SUDO="sudo -n"
+              ${SUDO} systemctl stop rustfs rustfs-rep2 2>/dev/null || true
+              if ${SUDO} dpkg -l rustfs 2>/dev/null | grep -q "^ii"; then
+                ${SUDO} dpkg -P rustfs
+              fi
+              for i in 1 2 3 4; do ${SUDO} rm -rf /data/rustfs${i}/mnmd; done
+              ${SUDO} rm -rf /data/rustfs-rep2 /var/log/rustfs /var/log/rustfs-rep2 /var/lib/rustfs/kms
+            '
+          done
+
+      - name: Run replication suite
+        id: test
+        continue-on-error: true
+        env:
+          LOG_FILE: /tmp/rustfs-replication.log
+        run: |
+          set -euo pipefail
+          chmod +x auto-testing/rustfs-replication-test.sh
+          PACKAGE_URL='${{ inputs.package_url }}'
+          RUSTFS_VERSION='${{ inputs.rustfs_version }}'
+          SUITE='${{ inputs.suite }}'
+          ARGS=(-y --log-file "${LOG_FILE}")
+          if [ "${SUITE}" = "all" ] || [ -z "${SUITE}" ] || [ "${SUITE}" = "null" ]; then
+            ARGS+=(--suite all)
+          else
+            ARGS+=(--suite "${SUITE}")
+          fi
+          if [ -n "${PACKAGE_URL}" ]; then
+            ARGS+=(--package-url "${PACKAGE_URL}")
+          elif [ -n "${RUSTFS_VERSION}" ] && [ "${RUSTFS_VERSION}" != "null" ]; then
+            ARGS+=(--version "${RUSTFS_VERSION}")
+          else
+            ARGS+=(--package-url "${RUSTFS_NIGHTLY_PACKAGE_URL}")
+          fi
+          ./auto-testing/rustfs-replication-test.sh "${ARGS[@]}"
+
+      - name: Generate report
+        if: always()
+        env:
+          LOG_FILE: /tmp/rustfs-replication.log
+          REPORT_FILE: /tmp/rustfs-replication-report.md
+        run: |
+          set -euo pipefail
+          PACKAGE_URL='${{ inputs.package_url }}'
+          RUSTFS_VERSION='${{ inputs.rustfs_version }}'
+          if [ -n "${PACKAGE_URL}" ]; then
+            PACKAGE_SOURCE="${PACKAGE_URL}"
+          elif [ -n "${RUSTFS_VERSION}" ] && [ "${RUSTFS_VERSION}" != "null" ]; then
+            PACKAGE_SOURCE="version ${RUSTFS_VERSION}"
+          else
+            PACKAGE_SOURCE="${RUSTFS_NIGHTLY_PACKAGE_URL}"
+          fi
+          read -r -a NODES <<< "${RUSTFS_NODES:-vm000 vm001 vm002}"
+          SSH_USER="${RUSTFS_SSH_USER:-azureuser}"
+          RUSTFS_VERSION_INFO="N/A"
+          if [ "${#NODES[@]}" -gt 0 ]; then
+            DETECTED_VERSION="$(ssh -o BatchMode=yes -o ConnectTimeout=10 -o StrictHostKeyChecking=accept-new \
+              "${SSH_USER}@${NODES[0]}" 'rustfs --version' 2>/dev/null | tr -d '\r' | head -n 1 || true)"
+            if [ -n "${DETECTED_VERSION}" ]; then
+              RUSTFS_VERSION_INFO="${DETECTED_VERSION}"
+            fi
+          fi
+          CASE_TABLE="/tmp/rustfs-replication-cases.md"
+          python3 - "${LOG_FILE}" "${CASE_TABLE}" <<'PY'
+          import re
+          import sys
+
+          log_file, out_file = sys.argv[1], sys.argv[2]
+          ansi = re.compile(r'\x1b\[[0-9;]*m')
+          start_re = re.compile(r'^---\s+([A-Z0-9]+-[0-9]+)\s+(.+?)\s+---$')
+          done_re = re.compile(r'^\[(PASS|FAIL|UNSUPPORTED)\]\s+([A-Z0-9]+-[0-9]+)\b')
+
+          rows = []
+          index = {}
+          try:
+              with open(log_file, 'r', encoding='utf-8', errors='replace') as fh:
+                  for raw in fh:
+                      line = ansi.sub('', raw).strip()
+                      m = start_re.match(line)
+                      if m:
+                          case_id, name = m.group(1), m.group(2)
+                          if case_id not in index:
+                              index[case_id] = len(rows)
+                              rows.append([case_id, name, 'RUNNING'])
+                          continue
+                      m = done_re.match(line)
+                      if m:
+                          status, case_id = m.group(1), m.group(2)
+                          if case_id in index:
+                              rows[index[case_id]][2] = status
+                          else:
+                              rows.append([case_id, case_id, status])
+                              index[case_id] = len(rows) - 1
+          except FileNotFoundError:
+              rows = []
+
+          counts = {'PASS': 0, 'FAIL': 0, 'UNSUPPORTED': 0, 'RUNNING': 0}
+          for _, _, status in rows:
+              counts[status] = counts.get(status, 0) + 1
+
+          with open(out_file, 'w', encoding='utf-8') as out:
+              out.write('## Case Summary\n\n')
+              out.write(f"- Total: {len(rows)}\\n")
+              out.write(f"- PASS: {counts.get('PASS', 0)}\\n")
+              out.write(f"- FAIL: {counts.get('FAIL', 0)}\\n")
+              out.write(f"- UNSUPPORTED: {counts.get('UNSUPPORTED', 0)}\\n")
+              out.write('\\n')
+              out.write('| Case | Name | Status |\\n')
+              out.write('| --- | --- | --- |\\n')
+              for case_id, name, status in rows:
+                  out.write(f'| {case_id} | {name} | {status} |\\n')
+          PY
+          {
+            echo "# RustFS replication test report"
+            echo ""
+            echo "- Run: ${{ github.server_url }}/${{ github.repository }}/actions/runs/${{ github.run_id }}"
+            echo "- Trigger: ${{ github.event_name }}"
+            echo "- Package: ${PACKAGE_SOURCE}"
+            echo "- RustFS Version: ${RUSTFS_VERSION_INFO}"
+            echo "- Test Step Outcome: ${{ steps.test.outcome }}"
+            echo ""
+            cat "${CASE_TABLE}" || true
+            echo ""
+            echo "## Log tail"
+            echo '```text'
+            tail -n 200 "${LOG_FILE}" || true
+            echo '```'
+          } | tee "${REPORT_FILE}"
+          cat "${REPORT_FILE}" >> "${GITHUB_STEP_SUMMARY}"
+
+      - name: Upload functional report to dashboard
+        if: always()
+        continue-on-error: true
+        env:
+          GH_TOKEN: ${{ env.PF_TESTING_GH_TOKEN }}
+          REPORT_FILE: /tmp/rustfs-replication-report.md
+          SUITE: replication
+        run: |
+          set -euo pipefail
+          if [ -z "${GH_TOKEN:-}" ]; then
+            echo "PF_TESTING_GH_TOKEN is not configured; skipping dashboard upload"
+            exit 0
+          fi
+          DATE="$(date -u +%Y-%m-%d)"
+          REPORT_PATH="functional-reports/${SUITE}/${DATE}.md"
+          CONTENT="$(python3 -c 'import base64,sys;print(base64.b64encode(open(sys.argv[1],"rb").read()).decode())' "${REPORT_FILE}")"
+          SHA="$(gh api "repos/rustfs/dashboard/contents/${REPORT_PATH}" -q '.sha' 2>/dev/null || true)"
+          if [ -n "${SHA}" ]; then
+            jq -n --arg msg "report(${SUITE}): ${DATE}" --arg content "${CONTENT}" --arg sha "${SHA}" \
+              '{message:$msg, content:$content, sha:$sha}' \
+              | gh api --method PUT "repos/rustfs/dashboard/contents/${REPORT_PATH}" --input - >/dev/null
+          else
+            jq -n --arg msg "report(${SUITE}): ${DATE}" --arg content "${CONTENT}" \
+              '{message:$msg, content:$content}' \
+              | gh api --method PUT "repos/rustfs/dashboard/contents/${REPORT_PATH}" --input - >/dev/null
+          fi
+
+      - name: File failure issue in rustfs/backlog
+        if: ${{ always() && (failure() || steps.test.outcome == 'failure' || steps.test.outcome == 'cancelled') }}
+        continue-on-error: true
+        env:
+          GH_TOKEN: ${{ secrets.PF_TESTING_GH_TOKEN }}
+          SUITE: 'replication'
+          SUITE_LABEL: 'Replication (bucket + site)'
+          RUN_URL: ${{ github.server_url }}/${{ github.repository }}/actions/runs/${{ github.run_id }}
+          REPORT_FILE: '/tmp/rustfs-replication-report.md'
+          LOG_FILE: '/tmp/rustfs-replication.log'
+        run: |
+          set -euo pipefail
+          if [ -z "${GH_TOKEN:-}" ]; then
+            echo "PF_TESTING_GH_TOKEN is not configured; skipping backlog issue"
+            exit 0
+          fi
+          TITLE="[functional][${SUITE}] ${SUITE_LABEL} suite failed (run ${GITHUB_RUN_ID})"
+          EXISTING="$(gh issue list -R rustfs/backlog --state all \
+            --search "in:title \"run ${GITHUB_RUN_ID}\"" \
+            --json number --jq '.[].number' || true)"
+          if [ -n "${EXISTING}" ]; then
+            echo "backlog issue already exists for run ${GITHUB_RUN_ID}; skipping"
+            exit 0
+          fi
+          redact() {
+            sed -E \
+              -e 's/(RUSTFS_(ACCESS_KEY|SECRET_KEY)[=: ]+)[^[:space:]]+/\1[REDACTED]/Ig' \
+              -e 's/(Authorization:).*/\1 [REDACTED]/Ig' \
+              -e 's/(X-Amz-Signature=)[^&[:space:]]+/\1[REDACTED]/Ig' \
+              -e 's/^.*(password|secret|token)[=: ].*/[REDACTED SENSITIVE LINE]/Ig'
+          }
+          BODY_FILE="$(mktemp)"
+          {
+            echo "The **${SUITE_LABEL}** functional suite failed."
+            echo ""
+            echo "- Suite: \`${SUITE}\`"
+            echo "- Run: ${RUN_URL}"
+            echo "- Trigger: ${GITHUB_EVENT_NAME}"
+            echo "- Date: $(date -u +%Y-%m-%d)"
+            echo ""
+            echo "## Report (errors and symptoms)"
+            echo ""
+            if [ -s "${REPORT_FILE}" ]; then
+              redact < "${REPORT_FILE}"
+            elif [ -s "${LOG_FILE:-}" ]; then
+              echo "(report file missing; log tail below)"
+              echo ""
+              tail -n 200 "${LOG_FILE}" | redact
+            else
+              echo "(no report or log file was produced)"
+            fi
+          } | head -c 55000 > "${BODY_FILE}"
+          gh label create functional-test -R rustfs/backlog --color d73a4a 2>/dev/null || true
+          if ! gh issue create -R rustfs/backlog --title "${TITLE}" \
+              --body-file "${BODY_FILE}" --label functional-test; then
+            gh issue create -R rustfs/backlog --title "${TITLE}" --body-file "${BODY_FILE}"
+          fi
+          echo "filed backlog issue for suite ${SUITE}"
+
+      - name: Upload report and logs
+        if: always()
+        uses: actions/upload-artifact@b7c566a772e6b6bfb58ed0dc250532a479d7789f # v6
+        with:
+          name: rustfs-replication-${{ github.run_id }}
+          path: |
+            /tmp/rustfs-replication.log
+            /tmp/rustfs-replication-report.md
+          if-no-files-found: warn
+
+      - name: Cleanup environment (after)
+        if: always()
+        run: |
+          set -euo pipefail
+          read -r -a NODES <<< "${RUSTFS_NODES:-vm000 vm001 vm002}"
+          SSH_USER="${RUSTFS_SSH_USER:-azureuser}"
+          for node in "${NODES[@]}"; do
+            ssh -o BatchMode=yes -o ConnectTimeout=10 -o StrictHostKeyChecking=accept-new "${SSH_USER}@${node}" '
+              set -euo pipefail
+              SUDO=""; [ "$(id -u)" -ne 0 ] && SUDO="sudo -n"
+              ${SUDO} systemctl stop rustfs rustfs-rep2 2>/dev/null || true
+              if ${SUDO} dpkg -l rustfs 2>/dev/null | grep -q "^ii"; then
+                ${SUDO} dpkg -P rustfs
+              fi
+              for i in 1 2 3 4; do ${SUDO} rm -rf /data/rustfs${i}/mnmd; done
+              ${SUDO} rm -rf /data/rustfs-rep2 /var/log/rustfs /var/log/rustfs-rep2
+            '
+          done
+
+      - name: Chain complete
+        # Replication is the last link of the functional chain: nothing to
+        # dispatch after it. This step just records that the chain finished.
+        if: ${{ always() && github.event_name == 'repository_dispatch' }}
+        run: |
+          echo "Functional chain complete: replication (final suite) finished."
+          echo "from_suite=security trigger=${{ github.event_name }} outcome=${{ steps.test.outcome }}"
+
+      - name: Notify on failure
+        if: failure()
+        run: |
+          echo "RustFS replication suite failed"
+          echo "Package source: ${{ inputs.package_url || inputs.rustfs_version || 'nightly (R2 latest)' }}"
+          echo "See the uploaded report and log artifacts for details."
@@ -0,0 +1,374 @@
+name: RustFS S3 Compatibility Test
+
+on:
+  workflow_dispatch:
+    inputs:
+      rustfs_version:
+        description: 'RustFS release tag to test (e.g. 1.0.0-rc.4-preview.1)'
+        required: false
+        default: '1.0.0-rc.4-preview.1'
+      package_url:
+        description: 'Direct .deb URL (nightly/R2/dev). Overrides rustfs_version.'
+        required: false
+        type: string
+  repository_dispatch:
+    # Chain handoff: dispatched when the upgrade suite finishes.
+    types: [rustfs-chain-s3]
+
+permissions:
+  contents: read
+
+concurrency:
+  group: rustfs-shared-functional-tests
+  cancel-in-progress: false
+
+defaults:
+  run:
+    shell: bash
+
+env:
+  RUSTFS_ACCESS_KEY: ${{ secrets.RUSTFS_ACCESS_KEY }}
+  RUSTFS_SECRET_KEY: ${{ secrets.RUSTFS_SECRET_KEY }}
+  RUSTFS_NODES: ${{ secrets.RUSTFS_NODES || vars.RUSTFS_NODES }}
+  RUSTFS_SSH_USER: ${{ secrets.RUSTFS_SSH_USER || vars.RUSTFS_SSH_USER }}
+  RUSTFS_NIGHTLY_PACKAGE_URL: ${{ vars.RUSTFS_NIGHTLY_PACKAGE_URL || 'https://dl.rustfs.com/artifacts/rustfs/packages/nightly/rustfs-nightly-latest.deb' }}
+  PF_TESTING_GH_TOKEN: ${{ secrets.PF_TESTING_GH_TOKEN }}
+
+jobs:
+  s3-compat-test:
+    runs-on: smoke-testing
+    continue-on-error: true
+    timeout-minutes: 360
+    if: ${{ github.event_name == 'workflow_dispatch' || github.event_name == 'repository_dispatch' }}
+    steps:
+      # auto-testing is private: clone it with the dedicated PF token (not
+      # GITHUB_TOKEN) and retry transient GitHub/network failures.
+      - name: Checkout auto-testing scripts (with retry)
+        env:
+          GH_TOKEN: ${{ secrets.PF_TESTING_GH_TOKEN }}
+        run: |
+          set -euo pipefail
+          rm -rf auto-testing
+          for attempt in 1 2 3 4 5; do
+            if gh repo clone rustfs/auto-testing auto-testing -- --depth 1 --quiet; then
+              echo "auto-testing cloned (attempt ${attempt})"
+              exit 0
+            fi
+            rm -rf auto-testing
+            echo "clone attempt ${attempt} failed; retrying in $((attempt * 15))s" >&2
+            sleep $((attempt * 15))
+          done
+          echo "ERROR: unable to clone rustfs/auto-testing after 5 attempts" >&2
+          exit 1
+
+      - name: Show environment
+        run: |
+          uname -a
+          jq --version
+          openssl version
+          df -h /data | tail -1
+
+      - name: Cleanup environment (before)
+        run: |
+          set -euo pipefail
+          read -r -a NODES <<< "${RUSTFS_NODES:-vm000 vm001 vm002}"
+          SSH_USER="${RUSTFS_SSH_USER:-azureuser}"
+          for node in "${NODES[@]}"; do
+            ssh -o BatchMode=yes -o ConnectTimeout=10 -o StrictHostKeyChecking=accept-new "${SSH_USER}@${node}" '
+              set -euo pipefail
+              SUDO=""; [ "$(id -u)" -ne 0 ] && SUDO="sudo -n"
+              ${SUDO} systemctl stop rustfs 2>/dev/null || true
+              if ${SUDO} dpkg -l rustfs 2>/dev/null | grep -q "^ii"; then
+                ${SUDO} dpkg -P rustfs
+              fi
+              for i in 1 2 3 4; do ${SUDO} rm -rf /data/rustfs${i}/mnmd; done
+              ${SUDO} rm -rf /var/log/rustfs
+            '
+          done
+
+      - name: Run S3 compatibility suite
+        id: test
+        continue-on-error: true
+        env:
+          LOG_FILE: /tmp/rustfs-s3-compat.log
+        run: |
+          set -euo pipefail
+          chmod +x auto-testing/rustfs-s3-compat-test.sh
+          PACKAGE_URL='${{ inputs.package_url }}'
+          RUSTFS_VERSION='${{ inputs.rustfs_version }}'
+          ARGS=(--all-topologies -y --log-file "${LOG_FILE}")
+          if [ -n "${PACKAGE_URL}" ]; then
+            ARGS+=(--package-url "${PACKAGE_URL}")
+          elif [ -n "${RUSTFS_VERSION}" ]; then
+            ARGS+=(--version "${RUSTFS_VERSION}")
+          else
+            ARGS+=(--package-url "${RUSTFS_NIGHTLY_PACKAGE_URL}")
+          fi
+          ./auto-testing/rustfs-s3-compat-test.sh "${ARGS[@]}"
+
+      - name: Generate report
+        if: always()
+        env:
+          LOG_FILE: /tmp/rustfs-s3-compat.log
+          REPORT_FILE: /tmp/rustfs-s3-compat-report.md
+        run: |
+          set -euo pipefail
+          PACKAGE_URL='${{ inputs.package_url }}'
+          RUSTFS_VERSION='${{ inputs.rustfs_version }}'
+          if [ -n "${PACKAGE_URL}" ]; then
+            PACKAGE_SOURCE="${PACKAGE_URL}"
+          elif [ -n "${RUSTFS_VERSION}" ]; then
+            PACKAGE_SOURCE="version ${RUSTFS_VERSION}"
+          else
+            PACKAGE_SOURCE="${RUSTFS_NIGHTLY_PACKAGE_URL}"
+          fi
+          read -r -a NODES <<< "${RUSTFS_NODES:-vm000 vm001 vm002}"
+          SSH_USER="${RUSTFS_SSH_USER:-azureuser}"
+          RUSTFS_VERSION_INFO="N/A"
+          if [ "${#NODES[@]}" -gt 0 ]; then
+            DETECTED_VERSION="$(ssh -o BatchMode=yes -o ConnectTimeout=10 -o StrictHostKeyChecking=accept-new \
+              "${SSH_USER}@${NODES[0]}" 'rustfs --version' 2>/dev/null | tr -d '\r' | head -n 1 || true)"
+            if [ -n "${DETECTED_VERSION}" ]; then
+              RUSTFS_VERSION_INFO="${DETECTED_VERSION}"
+            fi
+          fi
+          CASE_TABLE="/tmp/rustfs-s3-compat-cases.md"
+          python3 - "${LOG_FILE}" "${CASE_TABLE}" <<'PY'
+          import re
+          import sys
+
+          log_file, out_file = sys.argv[1], sys.argv[2]
+          ansi = re.compile(r'\x1b\[[0-9;]*m')
+          start_re = re.compile(r'^---\s+([A-Z0-9]+-[0-9]+)\s+(.+?)\s+---$')
+          done_re = re.compile(r'^\[(PASS|FAIL|UNSUPPORTED)\]\s+([A-Z0-9]+-[0-9]+)\b')
+
+          rows = []
+          index = {}
+          current = None
+          try:
+            with open(log_file, 'r', encoding='utf-8', errors='replace') as fh:
+              for raw in fh:
+                line = ansi.sub('', raw).strip()
+                m = start_re.match(line)
+                if m:
+                  case_id, name = m.group(1), m.group(2)
+                  current = case_id
+                  if case_id not in index:
+                    index[case_id] = len(rows)
+                    rows.append([case_id, name, 'RUNNING'])
+                  continue
+                m = done_re.match(line)
+                if m:
+                  status, case_id = m.group(1), m.group(2)
+                  if case_id in index:
+                    rows[index[case_id]][2] = status
+                  else:
+                    rows.append([case_id, case_id, status])
+                    index[case_id] = len(rows) - 1
+                  current = None
+          except FileNotFoundError:
+            rows = []
+
+          counts = {'PASS': 0, 'FAIL': 0, 'UNSUPPORTED': 0, 'RUNNING': 0}
+          for _, _, status in rows:
+            counts[status] = counts.get(status, 0) + 1
+
+          with open(out_file, 'w', encoding='utf-8') as out:
+            out.write('## Case Summary\n\n')
+            out.write(f"- Total: {len(rows)}\\n")
+            out.write(f"- PASS: {counts.get('PASS', 0)}\\n")
+            out.write(f"- FAIL: {counts.get('FAIL', 0)}\\n")
+            out.write(f"- UNSUPPORTED: {counts.get('UNSUPPORTED', 0)}\\n")
+            out.write('\\n')
+            out.write('| Case | Name | Status |\\n')
+            out.write('| --- | --- | --- |\\n')
+            for case_id, name, status in rows:
+              out.write(f'| {case_id} | {name} | {status} |\\n')
+          PY
+          {
+            echo "# RustFS S3 compatibility test report"
+            echo ""
+            echo "- Run: ${{ github.server_url }}/${{ github.repository }}/actions/runs/${{ github.run_id }}"
+            echo "- Trigger: ${{ github.event_name }}"
+            echo "- Package: ${PACKAGE_SOURCE}"
+            echo "- RustFS Version: ${RUSTFS_VERSION_INFO}"
+            echo "- Test Step Outcome: ${{ steps.test.outcome }}"
+            echo ""
+            cat "${CASE_TABLE}" || true
+            echo ""
+            echo "## Log tail"
+            echo '```text'
+            tail -n 200 "${LOG_FILE}" || true
+            echo '```'
+          } | tee "${REPORT_FILE}"
+          cat "${REPORT_FILE}" >> "${GITHUB_STEP_SUMMARY}"
+
+      - name: Upload functional report to dashboard
+        if: always()
+        continue-on-error: true
+        env:
+          GH_TOKEN: ${{ env.PF_TESTING_GH_TOKEN }}
+          REPORT_FILE: /tmp/rustfs-s3-compat-report.md
+          SUITE: s3
+        run: |
+          set -euo pipefail
+          if [ -z "${GH_TOKEN:-}" ]; then
+            echo "PF_TESTING_GH_TOKEN is not configured; skipping dashboard upload"
+            exit 0
+          fi
+          DATE="$(date -u +%Y-%m-%d)"
+          REPORT_PATH="functional-reports/${SUITE}/${DATE}.md"
+          CONTENT="$(python3 -c 'import base64,sys;print(base64.b64encode(open(sys.argv[1],"rb").read()).decode())' "${REPORT_FILE}")"
+          SHA="$(gh api "repos/rustfs/dashboard/contents/${REPORT_PATH}" -q '.sha' 2>/dev/null || true)"
+          if [ -n "${SHA}" ]; then
+            jq -n --arg msg "report(${SUITE}): ${DATE}" --arg content "${CONTENT}" --arg sha "${SHA}" \
+              '{message:$msg, content:$content, sha:$sha}' \
+              | gh api --method PUT "repos/rustfs/dashboard/contents/${REPORT_PATH}" --input - >/dev/null
+          else
+            jq -n --arg msg "report(${SUITE}): ${DATE}" --arg content "${CONTENT}" \
+              '{message:$msg, content:$content}' \
+              | gh api --method PUT "repos/rustfs/dashboard/contents/${REPORT_PATH}" --input - >/dev/null
+          fi
+
+      - name: File failure issue in rustfs/backlog
+        if: ${{ always() && (failure() || steps.test.outcome == 'failure' || steps.test.outcome == 'cancelled') }}
+        continue-on-error: true
+        env:
+          GH_TOKEN: ${{ secrets.PF_TESTING_GH_TOKEN }}
+          SUITE: 's3'
+          SUITE_LABEL: 'S3 compatibility'
+          RUN_URL: ${{ github.server_url }}/${{ github.repository }}/actions/runs/${{ github.run_id }}
+          REPORT_FILE: '/tmp/rustfs-s3-compat-report.md'
+          LOG_FILE: '/tmp/rustfs-s3-compat.log'
+        run: |
+          set -euo pipefail
+          if [ -z "${GH_TOKEN:-}" ]; then
+            echo "PF_TESTING_GH_TOKEN is not configured; skipping backlog issue"
+            exit 0
+          fi
+          TITLE="[functional][${SUITE}] ${SUITE_LABEL} suite failed (run ${GITHUB_RUN_ID})"
+          EXISTING="$(gh issue list -R rustfs/backlog --state all \
+            --search "in:title \"run ${GITHUB_RUN_ID}\"" \
+            --json number --jq '.[].number' || true)"
+          if [ -n "${EXISTING}" ]; then
+            echo "backlog issue already exists for run ${GITHUB_RUN_ID}; skipping"
+            exit 0
+          fi
+          redact() {
+            sed -E \
+              -e 's/(RUSTFS_(ACCESS_KEY|SECRET_KEY)[=: ]+)[^[:space:]]+/\1[REDACTED]/Ig' \
+              -e 's/(Authorization:).*/\1 [REDACTED]/Ig' \
+              -e 's/(X-Amz-Signature=)[^&[:space:]]+/\1[REDACTED]/Ig' \
+              -e 's/^.*(password|secret|token)[=: ].*/[REDACTED SENSITIVE LINE]/Ig'
+          }
+          BODY_FILE="$(mktemp)"
+          {
+            echo "The **${SUITE_LABEL}** functional suite failed."
+            echo ""
+            echo "- Suite: \`${SUITE}\`"
+            echo "- Run: ${RUN_URL}"
+            echo "- Trigger: ${GITHUB_EVENT_NAME}"
+            echo "- Date: $(date -u +%Y-%m-%d)"
+            echo ""
+            echo "## Report (errors and symptoms)"
+            echo ""
+            if [ -s "${REPORT_FILE}" ]; then
+              redact < "${REPORT_FILE}"
+            elif [ -s "${LOG_FILE:-}" ]; then
+              echo "(report file missing; log tail below)"
+              echo ""
+              tail -n 200 "${LOG_FILE}" | redact
+            else
+              echo "(no report or log file was produced)"
+            fi
+          } | head -c 55000 > "${BODY_FILE}"
+          gh label create functional-test -R rustfs/backlog --color d73a4a 2>/dev/null || true
+          if ! gh issue create -R rustfs/backlog --title "${TITLE}" \
+              --body-file "${BODY_FILE}" --label functional-test; then
+            gh issue create -R rustfs/backlog --title "${TITLE}" --body-file "${BODY_FILE}"
+          fi
+          echo "filed backlog issue for suite ${SUITE}"
+
+      - name: Upload report and logs
+        if: always()
+        uses: actions/upload-artifact@b7c566a772e6b6bfb58ed0dc250532a479d7789f # v6
+        with:
+          name: rustfs-s3-compat-${{ github.run_id }}
+          path: |
+            /tmp/rustfs-s3-compat.log
+            /tmp/rustfs-s3-compat-report.md
+          if-no-files-found: warn
+
+      - name: Cleanup environment (after)
+        if: always()
+        run: |
+          set -euo pipefail
+          read -r -a NODES <<< "${RUSTFS_NODES:-vm000 vm001 vm002}"
+          SSH_USER="${RUSTFS_SSH_USER:-azureuser}"
+          for node in "${NODES[@]}"; do
+            ssh -o BatchMode=yes -o ConnectTimeout=10 -o StrictHostKeyChecking=accept-new "${SSH_USER}@${node}" '
+              set -euo pipefail
+              SUDO=""; [ "$(id -u)" -ne 0 ] && SUDO="sudo -n"
+              ${SUDO} systemctl stop rustfs 2>/dev/null || true
+              if ${SUDO} dpkg -l rustfs 2>/dev/null | grep -q "^ii"; then
+                ${SUDO} dpkg -P rustfs
+              fi
+              for i in 1 2 3 4; do ${SUDO} rm -rf /data/rustfs${i}/mnmd; done
+              ${SUDO} rm -rf /var/log/rustfs
+            '
+          done
+
+      - name: "Continue functional chain (next: KMS)"
+        # Only chain-triggered runs forward to the next suite; standalone
+        # workflow_dispatch runs stop after their own cleanup. A failed
+        # handoff must never pass silently: it retries, then files an alert
+        # issue in rustfs/backlog so a stalled chain is visible.
+        if: ${{ always() && github.event_name == 'repository_dispatch' }}
+        continue-on-error: true
+        env:
+          GH_TOKEN: ${{ secrets.PF_TESTING_GH_TOKEN }}
+        run: |
+          set -uo pipefail
+          if [ -z "${GH_TOKEN:-}" ]; then
+            echo "PF_TESTING_GH_TOKEN is not configured; cannot dispatch the next suite" >&2
+            exit 1
+          fi
+          DISPATCHED=0
+          for attempt in 1 2 3; do
+            if gh api --method POST repos/rustfs/rustfs/dispatches \
+              -f event_type='rustfs-chain-kms' \
+              -F 'client_payload[from_suite]=s3'; then
+              echo "dispatched next suite KMS (attempt ${attempt})"
+              DISPATCHED=1
+              break
+            fi
+            echo "dispatch attempt ${attempt} failed; retrying in ${attempt}0s" >&2
+            sleep "${attempt}0"
+          done
+          if [ "${DISPATCHED:-0}" -ne 1 ]; then
+            echo "ERROR: functional chain stalled: could not dispatch KMS after 3 attempts" >&2
+            TITLE="[functional][chain] stalled after s3 (run ${GITHUB_RUN_ID})"
+            BODY_FILE="$(mktemp)"
+            {
+              echo "The functional chain could not hand off from **s3** to **KMS** after 3 attempts."
+              echo ""
+              echo "- Failed suite job: ${GITHUB_SERVER_URL}/${GITHUB_REPOSITORY}/actions/runs/${GITHUB_RUN_ID}"
+              echo "- Expected next event: 'rustfs-chain-kms'"
+              echo "- Likely cause: PF_TESTING_GH_TOKEN lacks contents:write on rustfs/rustfs, or the GitHub API was unavailable."
+              echo "- Recovery: re-dispatch manually with"
+              FENCE="$(printf "\x60\x60\x60")"; echo "  ${FENCE}"
+              echo "  gh api --method POST repos/rustfs/rustfs/dispatches -f event_type='rustfs-chain-kms'"
+              FENCE="$(printf "\x60\x60\x60")"; echo "  ${FENCE}"
+            } > "${BODY_FILE}"
+            gh issue create -R rustfs/backlog --title "${TITLE}" \
+              --body-file "${BODY_FILE}" --label functional-test \
+              || gh issue create -R rustfs/backlog --title "${TITLE}" --body-file "${BODY_FILE}" \
+              || echo "could not file the stall alert issue either; check the token" >&2
+            exit 1
+          fi
+
+      - name: Notify on failure
+        if: failure()
+        run: |
+          echo "RustFS S3 compatibility suite failed"
+          echo "See the uploaded report and log artifacts for details."
@@ -0,0 +1,318 @@
+# Copyright 2024 RustFS Team
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#     http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+
+name: RustFS Security Test
+
+on:
+  workflow_dispatch:
+    inputs:
+      rustfs_version:
+        description: 'RustFS release tag to test (e.g. 1.0.0-rc.4-preview.1)'
+        required: false
+        default: '1.0.0-rc.4-preview.1'
+      package_url:
+        description: 'Direct .deb URL (nightly/R2/dev). Overrides rustfs_version.'
+        required: false
+        type: string
+      topology:
+        description: 'Topology to run (all = SNSD, SNMD, MNMD)'
+        type: choice
+        options:
+          - all
+          - single-single
+          - single-multi
+          - multi-multi
+        default: all
+      oidc_live:
+        description: 'Run the live Keycloak OIDC/SSO gate as part of the suite'
+        type: boolean
+        default: true
+      cleanup_before:
+        description: 'Reset the nodes before the test (DESTROYS existing data/config)'
+        type: boolean
+        default: true
+      cleanup_after:
+        description: 'Reset the nodes after the test (DESTROYS test data/config)'
+        type: boolean
+        default: true
+  repository_dispatch:
+    # Chain handoff: dispatched when the pool expansion suite finishes.
+    types: [rustfs-chain-security]
+
+permissions:
+  contents: read
+
+# The security suite uses the same shared VMs as the other functional tests,
+# so it must serialize with them instead of running in parallel.
+concurrency:
+  group: rustfs-shared-functional-tests
+  cancel-in-progress: false
+
+defaults:
+  run:
+    shell: bash
+
+env:
+  RUSTFS_ACCESS_KEY: ${{ secrets.RUSTFS_ACCESS_KEY }}
+  RUSTFS_SECRET_KEY: ${{ secrets.RUSTFS_SECRET_KEY }}
+  RUSTFS_NODES: ${{ secrets.RUSTFS_NODES || vars.RUSTFS_NODES }}
+  RUSTFS_SSH_USER: ${{ secrets.RUSTFS_SSH_USER || vars.RUSTFS_SSH_USER }}
+  RUSTFS_NIGHTLY_PACKAGE_URL: ${{ vars.RUSTFS_NIGHTLY_PACKAGE_URL || 'https://dl.rustfs.com/artifacts/rustfs/packages/nightly/rustfs-nightly-latest.deb' }}
+  PF_TESTING_GH_TOKEN: ${{ secrets.PF_TESTING_GH_TOKEN }}
+
+jobs:
+  security-test:
+    runs-on: smoke-testing
+    continue-on-error: true
+    timeout-minutes: 360
+    if: ${{ github.event_name == 'workflow_dispatch' || github.event_name == 'repository_dispatch' }}
+    steps:
+      # auto-testing is private: clone it with the dedicated PF token (not
+      # GITHUB_TOKEN) and retry transient GitHub/network failures.
+      - name: Checkout auto-testing scripts (with retry)
+        env:
+          GH_TOKEN: ${{ secrets.PF_TESTING_GH_TOKEN }}
+        run: |
+          set -euo pipefail
+          rm -rf auto-testing
+          for attempt in 1 2 3 4 5; do
+            if gh repo clone rustfs/auto-testing auto-testing -- --depth 1 --quiet; then
+              echo "auto-testing cloned (attempt ${attempt})"
+              exit 0
+            fi
+            rm -rf auto-testing
+            echo "clone attempt ${attempt} failed; retrying in $((attempt * 15))s" >&2
+            sleep $((attempt * 15))
+          done
+          echo "ERROR: unable to clone rustfs/auto-testing after 5 attempts" >&2
+          exit 1
+
+      - name: Checkout repository (for the OIDC live gate script)
+        uses: actions/checkout@9c091bb21b7c1c1d1991bb908d89e4e9dddfe3e0 # v7
+        with:
+          persist-credentials: false
+
+      - name: Show environment
+        run: |
+          uname -a
+          jq --version
+          openssl version
+          aws --version || true
+          docker --version || true
+          df -h /data | tail -1
+
+      - name: Cleanup environment (before)
+        if: ${{ inputs.cleanup_before != 'false' || github.event_name != 'workflow_dispatch' }}
+        run: |
+          set -euo pipefail
+          read -r -a NODES <<< "${RUSTFS_NODES:-vm000 vm001 vm002}"
+          SSH_USER="${RUSTFS_SSH_USER:-azureuser}"
+          for node in "${NODES[@]}"; do
+            ssh -o BatchMode=yes -o ConnectTimeout=10 -o StrictHostKeyChecking=accept-new "${SSH_USER}@${node}" '
+              set -euo pipefail
+              SUDO=""; [ "$(id -u)" -ne 0 ] && SUDO="sudo -n"
+              ${SUDO} systemctl stop rustfs 2>/dev/null || true
+              if ${SUDO} dpkg -l rustfs 2>/dev/null | grep -q "^ii"; then
+                ${SUDO} dpkg -P rustfs
+              fi
+              for i in 1 2 3 4; do ${SUDO} rm -rf /data/rustfs${i}/mnmd; done
+              ${SUDO} rm -rf /var/log/rustfs /var/lib/rustfs/kms
+            '
+          done
+
+      - name: Run security suite
+        id: test
+        continue-on-error: true
+        env:
+          REPORT_FILE: /tmp/rustfs-security-report.md
+          RUSTFS_SECURITY_OIDC_LIVE_SCRIPT: ${{ github.workspace }}/scripts/test/oidc_keycloak_live.sh
+        run: |
+          set -euo pipefail
+          chmod +x auto-testing/rustfs-security-test.sh
+          PACKAGE_URL='${{ inputs.package_url }}'
+          RUSTFS_VERSION='${{ inputs.rustfs_version }}'
+          TOPOLOGY='${{ inputs.topology }}'
+          ARGS=(-y)
+          if [ "${TOPOLOGY}" = "all" ] || [ -z "${TOPOLOGY}" ] || [ "${TOPOLOGY}" = "null" ]; then
+            ARGS+=(--all-topologies)
+          else
+            ARGS+=(--topology "${TOPOLOGY}")
+          fi
+          if [ "${{ inputs.oidc_live }}" = "true" ] || [ "${{ github.event_name }}" != "workflow_dispatch" ]; then
+            ARGS+=(--oidc-live)
+          fi
+          if [ -n "${PACKAGE_URL}" ]; then
+            ARGS+=(--package-url "${PACKAGE_URL}")
+          elif [ -n "${RUSTFS_VERSION}" ] && [ "${RUSTFS_VERSION}" != "null" ]; then
+            ARGS+=(--version "${RUSTFS_VERSION}")
+          else
+            ARGS+=(--package-url "${RUSTFS_NIGHTLY_PACKAGE_URL}")
+          fi
+          ./auto-testing/rustfs-security-test.sh "${ARGS[@]}"
+
+      - name: Generate report
+        if: always()
+        run: |
+          set -euo pipefail
+          if [ ! -f /tmp/rustfs-security-report.md ]; then
+            {
+              echo "# RustFS security test report"
+              echo ""
+              echo "- Run: ${{ github.server_url }}/${{ github.repository }}/actions/runs/${{ github.run_id }}"
+              echo "- Trigger: ${{ github.event_name }}"
+              echo "- Test Step Outcome: failure (suite did not produce a report)"
+            } > /tmp/rustfs-security-report.md
+          fi
+          cat /tmp/rustfs-security-report.md >> "${GITHUB_STEP_SUMMARY}"
+
+      - name: Upload functional report to dashboard
+        if: always()
+        continue-on-error: true
+        env:
+          GH_TOKEN: ${{ env.PF_TESTING_GH_TOKEN }}
+          REPORT_FILE: /tmp/rustfs-security-report.md
+          SUITE: security
+        run: |
+          set -euo pipefail
+          if [ -z "${GH_TOKEN:-}" ]; then
+            echo "PF_TESTING_GH_TOKEN is not configured; skipping dashboard upload"
+            exit 0
+          fi
+          DATE="$(date -u +%Y-%m-%d)"
+          REPORT_PATH="functional-reports/${SUITE}/${DATE}.md"
+          CONTENT="$(python3 -c 'import base64,sys;print(base64.b64encode(open(sys.argv[1],"rb").read()).decode())' "${REPORT_FILE}")"
+          SHA="$(gh api "repos/rustfs/dashboard/contents/${REPORT_PATH}" -q '.sha' 2>/dev/null || true)"
+          if [ -n "${SHA}" ]; then
+            jq -n --arg msg "report(${SUITE}): ${DATE}" --arg content "${CONTENT}" --arg sha "${SHA}" \
+              '{message:$msg, content:$content, sha:$sha}' \
+              | gh api --method PUT "repos/rustfs/dashboard/contents/${REPORT_PATH}" --input - >/dev/null
+          else
+            jq -n --arg msg "report(${SUITE}): ${DATE}" --arg content "${CONTENT}" \
+              '{message:$msg, content:$content}' \
+              | gh api --method PUT "repos/rustfs/dashboard/contents/${REPORT_PATH}" --input - >/dev/null
+          fi
+
+      - name: File failure issue in rustfs/backlog
+        if: ${{ always() && (failure() || steps.test.outcome == 'failure' || steps.test.outcome == 'cancelled') }}
+        continue-on-error: true
+        env:
+          GH_TOKEN: ${{ secrets.PF_TESTING_GH_TOKEN }}
+          SUITE: 'security'
+          SUITE_LABEL: 'Security'
+          RUN_URL: ${{ github.server_url }}/${{ github.repository }}/actions/runs/${{ github.run_id }}
+          REPORT_FILE: '/tmp/rustfs-security-report.md'
+          LOG_FILE: ''
+        run: |
+          set -euo pipefail
+          if [ -z "${GH_TOKEN:-}" ]; then
+            echo "PF_TESTING_GH_TOKEN is not configured; skipping backlog issue"
+            exit 0
+          fi
+          TITLE="[functional][${SUITE}] ${SUITE_LABEL} suite failed (run ${GITHUB_RUN_ID})"
+          EXISTING="$(gh issue list -R rustfs/backlog --state all \
+            --search "in:title \"run ${GITHUB_RUN_ID}\"" \
+            --json number --jq '.[].number' || true)"
+          if [ -n "${EXISTING}" ]; then
+            echo "backlog issue already exists for run ${GITHUB_RUN_ID}; skipping"
+            exit 0
+          fi
+          redact() {
+            sed -E \
+              -e 's/(RUSTFS_(ACCESS_KEY|SECRET_KEY)[=: ]+)[^[:space:]]+/\1[REDACTED]/Ig' \
+              -e 's/(Authorization:).*/\1 [REDACTED]/Ig' \
+              -e 's/(X-Amz-Signature=)[^&[:space:]]+/\1[REDACTED]/Ig' \
+              -e 's/^.*(password|secret|token)[=: ].*/[REDACTED SENSITIVE LINE]/Ig'
+          }
+          BODY_FILE="$(mktemp)"
+          {
+            echo "The **${SUITE_LABEL}** functional suite failed."
+            echo ""
+            echo "- Suite: \`${SUITE}\`"
+            echo "- Run: ${RUN_URL}"
+            echo "- Trigger: ${GITHUB_EVENT_NAME}"
+            echo "- Date: $(date -u +%Y-%m-%d)"
+            echo ""
+            echo "## Report (errors and symptoms)"
+            echo ""
+            if [ -s "${REPORT_FILE}" ]; then
+              redact < "${REPORT_FILE}"
+            elif [ -s "${LOG_FILE:-}" ]; then
+              echo "(report file missing; log tail below)"
+              echo ""
+              tail -n 200 "${LOG_FILE}" | redact
+            else
+              echo "(no report or log file was produced)"
+            fi
+          } | head -c 55000 > "${BODY_FILE}"
+          gh label create functional-test -R rustfs/backlog --color d73a4a 2>/dev/null || true
+          if ! gh issue create -R rustfs/backlog --title "${TITLE}" \
+              --body-file "${BODY_FILE}" --label functional-test; then
+            gh issue create -R rustfs/backlog --title "${TITLE}" --body-file "${BODY_FILE}"
+          fi
+          echo "filed backlog issue for suite ${SUITE}"
+
+      - name: Upload report and logs
+        if: always()
+        uses: actions/upload-artifact@b7c566a772e6b6bfb58ed0dc250532a479d7789f # v6
+        with:
+          name: rustfs-security-test-${{ github.run_id }}
+          path: |
+            /tmp/rustfs-security-report.md
+            /tmp/rustfs-security.*/*
+          if-no-files-found: ignore
+          retention-days: 3
+
+      - name: Cleanup environment (after)
+        if: ${{ always() && (inputs.cleanup_after != 'false' || github.event_name != 'workflow_dispatch') }}
+        run: |
+          set -euo pipefail
+          read -r -a NODES <<< "${RUSTFS_NODES:-vm000 vm001 vm002}"
+          SSH_USER="${RUSTFS_SSH_USER:-azureuser}"
+          for node in "${NODES[@]}"; do
+            ssh -o BatchMode=yes -o ConnectTimeout=10 -o StrictHostKeyChecking=accept-new "${SSH_USER}@${node}" '
+              set -euo pipefail
+              SUDO=""; [ "$(id -u)" -ne 0 ] && SUDO="sudo -n"
+              ${SUDO} systemctl stop rustfs 2>/dev/null || true
+              if ${SUDO} dpkg -l rustfs 2>/dev/null | grep -q "^ii"; then
+                ${SUDO} dpkg -P rustfs
+              fi
+              for i in 1 2 3 4; do ${SUDO} rm -rf /data/rustfs${i}/mnmd; done
+              ${SUDO} rm -rf /var/log/rustfs /var/lib/rustfs/kms
+            '
+          done
+
+      - name: "Continue functional chain (next: Replication)"
+        # Only chain-triggered runs forward to the next suite; standalone
+        # workflow_dispatch runs stop after their own cleanup.
+        if: ${{ always() && github.event_name == 'repository_dispatch' }}
+        continue-on-error: true
+        env:
+          GH_TOKEN: ${{ secrets.PF_TESTING_GH_TOKEN }}
+        run: |
+          set -euo pipefail
+          if [ -z "${GH_TOKEN:-}" ]; then
+            echo "PF_TESTING_GH_TOKEN is not configured; cannot dispatch the next suite" >&2
+            exit 1
+          fi
+          echo "Dispatching next functional suite: Replication"
+          gh api --method POST repos/rustfs/rustfs/dispatches \
+            -f event_type='rustfs-chain-replication' \
+            -F 'client_payload[from_suite]=security'
+
+      - name: Notify on failure
+        if: failure()
+        run: |
+          echo "RustFS security test failed"
+          echo "Package source: ${{ inputs.package_url || 'nightly (R2 latest)' }}"
+          echo "See the uploaded report and logs for details."
@@ -0,0 +1,389 @@
+name: RustFS Storage Engine Test
+
+on:
+  workflow_dispatch:
+    inputs:
+      rustfs_version:
+        description: 'RustFS release tag to test (e.g. 1.0.0-rc.4-preview.1)'
+        required: false
+        default: '1.0.0-rc.4-preview.1'
+      package_url:
+        description: 'Direct .deb URL (nightly/R2/dev). Overrides rustfs_version.'
+        required: false
+        type: string
+      topology:
+        description: 'Topology to run (all = SNSD, SNMD, MNMD)'
+        type: choice
+        options:
+          - all
+          - single-single
+          - single-multi
+          - multi-multi
+        default: all
+  repository_dispatch:
+    # Chain handoff: dispatched when the tier suite finishes.
+    types: [rustfs-chain-storage]
+
+permissions:
+  contents: read
+
+concurrency:
+  group: rustfs-shared-functional-tests
+  cancel-in-progress: false
+
+defaults:
+  run:
+    shell: bash
+
+env:
+  RUSTFS_ACCESS_KEY: ${{ secrets.RUSTFS_ACCESS_KEY }}
+  RUSTFS_SECRET_KEY: ${{ secrets.RUSTFS_SECRET_KEY }}
+  RUSTFS_NODES: ${{ secrets.RUSTFS_NODES || vars.RUSTFS_NODES }}
+  RUSTFS_SSH_USER: ${{ secrets.RUSTFS_SSH_USER || vars.RUSTFS_SSH_USER }}
+  RUSTFS_NIGHTLY_PACKAGE_URL: ${{ vars.RUSTFS_NIGHTLY_PACKAGE_URL || 'https://dl.rustfs.com/artifacts/rustfs/packages/nightly/rustfs-nightly-latest.deb' }}
+  PF_TESTING_GH_TOKEN: ${{ secrets.PF_TESTING_GH_TOKEN }}
+
+jobs:
+  storage-test:
+    runs-on: smoke-testing
+    continue-on-error: true
+    timeout-minutes: 360
+    if: ${{ github.event_name == 'workflow_dispatch' || github.event_name == 'repository_dispatch' }}
+    steps:
+      # auto-testing is private: clone it with the dedicated PF token (not
+      # GITHUB_TOKEN) and retry transient GitHub/network failures.
+      - name: Checkout auto-testing scripts (with retry)
+        env:
+          GH_TOKEN: ${{ secrets.PF_TESTING_GH_TOKEN }}
+        run: |
+          set -euo pipefail
+          rm -rf auto-testing
+          for attempt in 1 2 3 4 5; do
+            if gh repo clone rustfs/auto-testing auto-testing -- --depth 1 --quiet; then
+              echo "auto-testing cloned (attempt ${attempt})"
+              exit 0
+            fi
+            rm -rf auto-testing
+            echo "clone attempt ${attempt} failed; retrying in $((attempt * 15))s" >&2
+            sleep $((attempt * 15))
+          done
+          echo "ERROR: unable to clone rustfs/auto-testing after 5 attempts" >&2
+          exit 1
+
+      - name: Show environment
+        run: |
+          uname -a
+          jq --version
+          openssl version
+          df -h /data | tail -1
+
+      - name: Cleanup environment (before)
+        run: |
+          set -euo pipefail
+          read -r -a NODES <<< "${RUSTFS_NODES:-vm000 vm001 vm002}"
+          SSH_USER="${RUSTFS_SSH_USER:-azureuser}"
+          for node in "${NODES[@]}"; do
+            ssh -o BatchMode=yes -o ConnectTimeout=10 -o StrictHostKeyChecking=accept-new "${SSH_USER}@${node}" '
+              set -euo pipefail
+              SUDO=""; [ "$(id -u)" -ne 0 ] && SUDO="sudo -n"
+              ${SUDO} systemctl stop rustfs 2>/dev/null || true
+              if ${SUDO} dpkg -l rustfs 2>/dev/null | grep -q "^ii"; then
+                ${SUDO} dpkg -P rustfs
+              fi
+              for i in 1 2 3 4; do ${SUDO} rm -rf /data/rustfs${i}/mnmd; done
+              ${SUDO} rm -rf /var/log/rustfs /var/lib/rustfs/kms /var/lib/rustfs/kms-backup
+            '
+          done
+
+      - name: Run storage engine suite
+        id: test
+        continue-on-error: true
+        env:
+          LOG_FILE: /tmp/rustfs-storage.log
+        run: |
+          set -euo pipefail
+          chmod +x auto-testing/rustfs-storage-test.sh
+          PACKAGE_URL='${{ inputs.package_url }}'
+          RUSTFS_VERSION='${{ inputs.rustfs_version }}'
+          TOPOLOGY='${{ inputs.topology }}'
+          ARGS=(-y --log-file "${LOG_FILE}")
+          if [ "${TOPOLOGY}" = "all" ] || [ -z "${TOPOLOGY}" ] || [ "${TOPOLOGY}" = "null" ]; then
+            ARGS+=(--all-topologies)
+          else
+            ARGS+=(--topology "${TOPOLOGY}")
+          fi
+          if [ -n "${PACKAGE_URL}" ]; then
+            ARGS+=(--package-url "${PACKAGE_URL}")
+          elif [ -n "${RUSTFS_VERSION}" ]; then
+            ARGS+=(--version "${RUSTFS_VERSION}")
+          else
+            ARGS+=(--package-url "${RUSTFS_NIGHTLY_PACKAGE_URL}")
+          fi
+          ./auto-testing/rustfs-storage-test.sh "${ARGS[@]}"
+
+      - name: Generate report
+        if: always()
+        env:
+          LOG_FILE: /tmp/rustfs-storage.log
+          REPORT_FILE: /tmp/rustfs-storage-report.md
+        run: |
+          set -euo pipefail
+          PACKAGE_URL='${{ inputs.package_url }}'
+          RUSTFS_VERSION='${{ inputs.rustfs_version }}'
+          if [ -n "${PACKAGE_URL}" ]; then
+            PACKAGE_SOURCE="${PACKAGE_URL}"
+          elif [ -n "${RUSTFS_VERSION}" ]; then
+            PACKAGE_SOURCE="version ${RUSTFS_VERSION}"
+          else
+            PACKAGE_SOURCE="${RUSTFS_NIGHTLY_PACKAGE_URL}"
+          fi
+          read -r -a NODES <<< "${RUSTFS_NODES:-vm000 vm001 vm002}"
+          SSH_USER="${RUSTFS_SSH_USER:-azureuser}"
+          RUSTFS_VERSION_INFO="N/A"
+          if [ "${#NODES[@]}" -gt 0 ]; then
+            DETECTED_VERSION="$(ssh -o BatchMode=yes -o ConnectTimeout=10 -o StrictHostKeyChecking=accept-new \
+              "${SSH_USER}@${NODES[0]}" 'rustfs --version' 2>/dev/null | tr -d '\r' | head -n 1 || true)"
+            if [ -n "${DETECTED_VERSION}" ]; then
+              RUSTFS_VERSION_INFO="${DETECTED_VERSION}"
+            fi
+          fi
+          CASE_TABLE="/tmp/rustfs-storage-cases.md"
+          python3 - "${LOG_FILE}" "${CASE_TABLE}" <<'PY'
+          import re
+          import sys
+
+          log_file, out_file = sys.argv[1], sys.argv[2]
+          ansi = re.compile(r'\x1b\[[0-9;]*m')
+          start_re = re.compile(r'^---\s+([A-Z0-9]+-[0-9]+)\s+(.+?)\s+---$')
+          done_re = re.compile(r'^\[(PASS|FAIL|UNSUPPORTED)\]\s+([A-Z0-9]+-[0-9]+)\b')
+
+          rows = []
+          index = {}
+          current = None
+          try:
+            with open(log_file, 'r', encoding='utf-8', errors='replace') as fh:
+              for raw in fh:
+                line = ansi.sub('', raw).strip()
+                m = start_re.match(line)
+                if m:
+                  case_id, name = m.group(1), m.group(2)
+                  current = case_id
+                  if case_id not in index:
+                    index[case_id] = len(rows)
+                    rows.append([case_id, name, 'RUNNING'])
+                  continue
+                m = done_re.match(line)
+                if m:
+                  status, case_id = m.group(1), m.group(2)
+                  if case_id in index:
+                    rows[index[case_id]][2] = status
+                  else:
+                    rows.append([case_id, case_id, status])
+                    index[case_id] = len(rows) - 1
+                  current = None
+          except FileNotFoundError:
+            rows = []
+
+          counts = {'PASS': 0, 'FAIL': 0, 'UNSUPPORTED': 0, 'RUNNING': 0}
+          for _, _, status in rows:
+            counts[status] = counts.get(status, 0) + 1
+
+          with open(out_file, 'w', encoding='utf-8') as out:
+            out.write('## Case Summary\n\n')
+            out.write(f"- Total: {len(rows)}\\n")
+            out.write(f"- PASS: {counts.get('PASS', 0)}\\n")
+            out.write(f"- FAIL: {counts.get('FAIL', 0)}\\n")
+            out.write(f"- UNSUPPORTED: {counts.get('UNSUPPORTED', 0)}\\n")
+            out.write('\\n')
+            out.write('| Case | Name | Status |\\n')
+            out.write('| --- | --- | --- |\\n')
+            for case_id, name, status in rows:
+              out.write(f'| {case_id} | {name} | {status} |\\n')
+          PY
+          {
+            echo "# RustFS storage engine test report"
+            echo ""
+            echo "- Run: ${{ github.server_url }}/${{ github.repository }}/actions/runs/${{ github.run_id }}"
+            echo "- Trigger: ${{ github.event_name }}"
+            echo "- Package: ${PACKAGE_SOURCE}"
+            echo "- RustFS Version: ${RUSTFS_VERSION_INFO}"
+            echo "- Test Step Outcome: ${{ steps.test.outcome }}"
+            echo ""
+            cat "${CASE_TABLE}" || true
+            echo ""
+            echo "## Log tail"
+            echo '```text'
+            tail -n 200 "${LOG_FILE}" || true
+            echo '```'
+          } | tee "${REPORT_FILE}"
+          cat "${REPORT_FILE}" >> "${GITHUB_STEP_SUMMARY}"
+
+      - name: Upload functional report to dashboard
+        if: always()
+        continue-on-error: true
+        env:
+          GH_TOKEN: ${{ env.PF_TESTING_GH_TOKEN }}
+          REPORT_FILE: /tmp/rustfs-storage-report.md
+          SUITE: storage
+        run: |
+          set -euo pipefail
+          if [ -z "${GH_TOKEN:-}" ]; then
+            echo "PF_TESTING_GH_TOKEN is not configured; skipping dashboard upload"
+            exit 0
+          fi
+          DATE="$(date -u +%Y-%m-%d)"
+          REPORT_PATH="functional-reports/${SUITE}/${DATE}.md"
+          CONTENT="$(python3 -c 'import base64,sys;print(base64.b64encode(open(sys.argv[1],"rb").read()).decode())' "${REPORT_FILE}")"
+          SHA="$(gh api "repos/rustfs/dashboard/contents/${REPORT_PATH}" -q '.sha' 2>/dev/null || true)"
+          if [ -n "${SHA}" ]; then
+            jq -n --arg msg "report(${SUITE}): ${DATE}" --arg content "${CONTENT}" --arg sha "${SHA}" \
+              '{message:$msg, content:$content, sha:$sha}' \
+              | gh api --method PUT "repos/rustfs/dashboard/contents/${REPORT_PATH}" --input - >/dev/null
+          else
+            jq -n --arg msg "report(${SUITE}): ${DATE}" --arg content "${CONTENT}" \
+              '{message:$msg, content:$content}' \
+              | gh api --method PUT "repos/rustfs/dashboard/contents/${REPORT_PATH}" --input - >/dev/null
+          fi
+
+      - name: File failure issue in rustfs/backlog
+        if: ${{ always() && (failure() || steps.test.outcome == 'failure' || steps.test.outcome == 'cancelled') }}
+        continue-on-error: true
+        env:
+          GH_TOKEN: ${{ secrets.PF_TESTING_GH_TOKEN }}
+          SUITE: 'storage'
+          SUITE_LABEL: 'Storage engine'
+          RUN_URL: ${{ github.server_url }}/${{ github.repository }}/actions/runs/${{ github.run_id }}
+          REPORT_FILE: '/tmp/rustfs-storage-report.md'
+          LOG_FILE: '/tmp/rustfs-storage.log'
+        run: |
+          set -euo pipefail
+          if [ -z "${GH_TOKEN:-}" ]; then
+            echo "PF_TESTING_GH_TOKEN is not configured; skipping backlog issue"
+            exit 0
+          fi
+          TITLE="[functional][${SUITE}] ${SUITE_LABEL} suite failed (run ${GITHUB_RUN_ID})"
+          EXISTING="$(gh issue list -R rustfs/backlog --state all \
+            --search "in:title \"run ${GITHUB_RUN_ID}\"" \
+            --json number --jq '.[].number' || true)"
+          if [ -n "${EXISTING}" ]; then
+            echo "backlog issue already exists for run ${GITHUB_RUN_ID}; skipping"
+            exit 0
+          fi
+          redact() {
+            sed -E \
+              -e 's/(RUSTFS_(ACCESS_KEY|SECRET_KEY)[=: ]+)[^[:space:]]+/\1[REDACTED]/Ig' \
+              -e 's/(Authorization:).*/\1 [REDACTED]/Ig' \
+              -e 's/(X-Amz-Signature=)[^&[:space:]]+/\1[REDACTED]/Ig' \
+              -e 's/^.*(password|secret|token)[=: ].*/[REDACTED SENSITIVE LINE]/Ig'
+          }
+          BODY_FILE="$(mktemp)"
+          {
+            echo "The **${SUITE_LABEL}** functional suite failed."
+            echo ""
+            echo "- Suite: \`${SUITE}\`"
+            echo "- Run: ${RUN_URL}"
+            echo "- Trigger: ${GITHUB_EVENT_NAME}"
+            echo "- Date: $(date -u +%Y-%m-%d)"
+            echo ""
+            echo "## Report (errors and symptoms)"
+            echo ""
+            if [ -s "${REPORT_FILE}" ]; then
+              redact < "${REPORT_FILE}"
+            elif [ -s "${LOG_FILE:-}" ]; then
+              echo "(report file missing; log tail below)"
+              echo ""
+              tail -n 200 "${LOG_FILE}" | redact
+            else
+              echo "(no report or log file was produced)"
+            fi
+          } | head -c 55000 > "${BODY_FILE}"
+          gh label create functional-test -R rustfs/backlog --color d73a4a 2>/dev/null || true
+          if ! gh issue create -R rustfs/backlog --title "${TITLE}" \
+              --body-file "${BODY_FILE}" --label functional-test; then
+            gh issue create -R rustfs/backlog --title "${TITLE}" --body-file "${BODY_FILE}"
+          fi
+          echo "filed backlog issue for suite ${SUITE}"
+
+      - name: Upload report and logs
+        if: always()
+        uses: actions/upload-artifact@b7c566a772e6b6bfb58ed0dc250532a479d7789f # v6
+        with:
+          name: rustfs-storage-${{ github.run_id }}
+          path: |
+            /tmp/rustfs-storage.log
+            /tmp/rustfs-storage-report.md
+          if-no-files-found: warn
+
+      - name: Cleanup environment (after)
+        if: always()
+        run: |
+          set -euo pipefail
+          read -r -a NODES <<< "${RUSTFS_NODES:-vm000 vm001 vm002}"
+          SSH_USER="${RUSTFS_SSH_USER:-azureuser}"
+          for node in "${NODES[@]}"; do
+            ssh -o BatchMode=yes -o ConnectTimeout=10 -o StrictHostKeyChecking=accept-new "${SSH_USER}@${node}" '
+              set -euo pipefail
+              SUDO=""; [ "$(id -u)" -ne 0 ] && SUDO="sudo -n"
+              ${SUDO} systemctl stop rustfs 2>/dev/null || true
+              if ${SUDO} dpkg -l rustfs 2>/dev/null | grep -q "^ii"; then
+                ${SUDO} dpkg -P rustfs
+              fi
+              for i in 1 2 3 4; do ${SUDO} rm -rf /data/rustfs${i}/mnmd; done
+              ${SUDO} rm -rf /var/log/rustfs
+            '
+          done
+
+      - name: "Continue functional chain (next: Heal)"
+        # Only chain-triggered runs forward to the next suite; standalone
+        # workflow_dispatch runs stop after their own cleanup. A failed
+        # handoff must never pass silently: it retries, then files an alert
+        # issue in rustfs/backlog so a stalled chain is visible.
+        if: ${{ always() && github.event_name == 'repository_dispatch' }}
+        continue-on-error: true
+        env:
+          GH_TOKEN: ${{ secrets.PF_TESTING_GH_TOKEN }}
+        run: |
+          set -uo pipefail
+          if [ -z "${GH_TOKEN:-}" ]; then
+            echo "PF_TESTING_GH_TOKEN is not configured; cannot dispatch the next suite" >&2
+            exit 1
+          fi
+          DISPATCHED=0
+          for attempt in 1 2 3; do
+            if gh api --method POST repos/rustfs/rustfs/dispatches \
+              -f event_type='rustfs-chain-heal' \
+              -F 'client_payload[from_suite]=storage'; then
+              echo "dispatched next suite Heal (attempt ${attempt})"
+              DISPATCHED=1
+              break
+            fi
+            echo "dispatch attempt ${attempt} failed; retrying in ${attempt}0s" >&2
+            sleep "${attempt}0"
+          done
+          if [ "${DISPATCHED:-0}" -ne 1 ]; then
+            echo "ERROR: functional chain stalled: could not dispatch Heal after 3 attempts" >&2
+            TITLE="[functional][chain] stalled after storage (run ${GITHUB_RUN_ID})"
+            BODY_FILE="$(mktemp)"
+            {
+              echo "The functional chain could not hand off from **storage** to **Heal** after 3 attempts."
+              echo ""
+              echo "- Failed suite job: ${GITHUB_SERVER_URL}/${GITHUB_REPOSITORY}/actions/runs/${GITHUB_RUN_ID}"
+              echo "- Expected next event: 'rustfs-chain-heal'"
+              echo "- Likely cause: PF_TESTING_GH_TOKEN lacks contents:write on rustfs/rustfs, or the GitHub API was unavailable."
+              echo "- Recovery: re-dispatch manually with"
+              FENCE="$(printf "\x60\x60\x60")"; echo "  ${FENCE}"
+              echo "  gh api --method POST repos/rustfs/rustfs/dispatches -f event_type='rustfs-chain-heal'"
+              FENCE="$(printf "\x60\x60\x60")"; echo "  ${FENCE}"
+            } > "${BODY_FILE}"
+            gh issue create -R rustfs/backlog --title "${TITLE}" \
+              --body-file "${BODY_FILE}" --label functional-test \
+              || gh issue create -R rustfs/backlog --title "${TITLE}" --body-file "${BODY_FILE}" \
+              || echo "could not file the stall alert issue either; check the token" >&2
+            exit 1
+          fi
+
+      - name: Notify on failure
+        if: failure()
+        run: |
+          echo "RustFS storage engine suite failed"
+          echo "See the uploaded report and log artifacts for details."
@@ -0,0 +1,484 @@
+name: RustFS Tier Test
+
+on:
+  workflow_dispatch:
+    inputs:
+      rustfs_version:
+        description: 'RustFS release tag to test (e.g. 1.0.0-rc.4-preview.1)'
+        required: false
+        default: '1.0.0-rc.4-preview.1'
+      package_url:
+        description: 'Direct .deb URL (nightly/R2/dev). Overrides rustfs_version.'
+        required: false
+        type: string
+      rc_sha256:
+        description: 'Optional SHA-256 for the preinstalled rc binary; mismatch is an infrastructure failure.'
+        required: false
+        type: string
+      force_case_failure:
+        description: 'Diagnostic only: rewrite single-single/TIER-101 to FAIL after execution to verify artifact and final-gate behavior.'
+        required: false
+        default: false
+        type: boolean
+  repository_dispatch:
+    # Chain handoff: dispatched when the KMS suite finishes.
+    types: [rustfs-chain-tier]
+
+permissions:
+  contents: read
+
+concurrency:
+  group: rustfs-shared-functional-tests
+  cancel-in-progress: false
+
+defaults:
+  run:
+    shell: bash
+
+env:
+  RUSTFS_ACCESS_KEY: ${{ secrets.RUSTFS_ACCESS_KEY }}
+  RUSTFS_SECRET_KEY: ${{ secrets.RUSTFS_SECRET_KEY }}
+  RUSTFS_NODES: ${{ secrets.RUSTFS_NODES || vars.RUSTFS_NODES }}
+  RUSTFS_SSH_USER: ${{ secrets.RUSTFS_SSH_USER || vars.RUSTFS_SSH_USER }}
+  RUSTFS_NIGHTLY_PACKAGE_URL: ${{ vars.RUSTFS_NIGHTLY_PACKAGE_URL || 'https://dl.rustfs.com/artifacts/rustfs/packages/nightly/rustfs-nightly-latest.deb' }}
+  RUSTFS_EXPECTED_RC_SHA256: ${{ inputs.rc_sha256 || vars.RUSTFS_TIER_RC_SHA256 }}
+  PF_TESTING_GH_TOKEN: ${{ secrets.PF_TESTING_GH_TOKEN }}
+  TIER_ARTIFACTS_DIR: /tmp/rustfs-tier-artifacts-${{ github.run_id }}-${{ github.run_attempt }}
+
+jobs:
+  tier-test:
+    runs-on: smoke-testing
+    # Requirement: a failing suite must not fail the workflow; failures
+    # are filed to rustfs/backlog and the chain continues.
+    continue-on-error: true
+    timeout-minutes: 420
+    if: ${{ github.event_name == 'workflow_dispatch' || github.event_name == 'repository_dispatch' }}
+    steps:
+      - name: Initialize run evidence directory
+        id: evidence
+        run: |
+          set -euo pipefail
+          umask 077
+          if ! mkdir -- "${TIER_ARTIFACTS_DIR}"; then
+            echo "refusing to reuse tier evidence path: ${TIER_ARTIFACTS_DIR}" >&2
+            exit 1
+          fi
+          test -d "${TIER_ARTIFACTS_DIR}"
+          test ! -L "${TIER_ARTIFACTS_DIR}"
+
+      # auto-testing is private: clone it with the dedicated PF token (not
+      # GITHUB_TOKEN) and retry transient GitHub/network failures.
+      - name: Checkout auto-testing scripts (with retry)
+        env:
+          GH_TOKEN: ${{ secrets.PF_TESTING_GH_TOKEN }}
+        run: |
+          set -euo pipefail
+          rm -rf auto-testing
+          for attempt in 1 2 3 4 5; do
+            if gh repo clone rustfs/auto-testing auto-testing -- --depth 1 --quiet; then
+              echo "auto-testing cloned (attempt ${attempt})"
+              exit 0
+            fi
+            rm -rf auto-testing
+            echo "clone attempt ${attempt} failed; retrying in $((attempt * 15))s" >&2
+            sleep $((attempt * 15))
+          done
+          echo "ERROR: unable to clone rustfs/auto-testing after 5 attempts" >&2
+          exit 1
+
+      - name: Show environment
+        run: |
+          uname -a
+          jq --version
+          openssl version
+          df -h /data | tail -1
+
+      - name: Cleanup environment (before)
+        run: |
+          set -euo pipefail
+          sudo docker rm -f rustfs-test-mqtt >/dev/null 2>&1 || true
+          sudo rm -f /tmp/rustfs-mosquitto.conf
+          read -r -a NODES <<< "${RUSTFS_NODES:-vm000 vm001 vm002}"
+          SSH_USER="${RUSTFS_SSH_USER:-azureuser}"
+          for node in "${NODES[@]}"; do
+            ssh -o BatchMode=yes -o ConnectTimeout=10 -o StrictHostKeyChecking=accept-new "${SSH_USER}@${node}" '
+              set -euo pipefail
+              SUDO=""; [ "$(id -u)" -ne 0 ] && SUDO="sudo -n"
+              ${SUDO} systemctl stop rustfs 2>/dev/null || true
+              if ${SUDO} dpkg -l rustfs 2>/dev/null | grep -q "^ii"; then
+                ${SUDO} dpkg -P rustfs
+              fi
+              for i in 1 2 3 4; do ${SUDO} rm -rf /data/rustfs${i}/mnmd; done
+              ${SUDO} rm -rf /var/log/rustfs /var/lib/rustfs/kms
+            '
+          done
+
+      - name: Ensure MQTT broker + clients
+        run: |
+          set -euo pipefail
+          if ! command -v mosquitto_sub >/dev/null 2>&1; then
+            sudo apt-get update
+            sudo apt-get install -y mosquitto-clients
+          fi
+          command -v docker >/dev/null 2>&1 || { echo 'docker not found on runner'; exit 1; }
+          sudo docker rm -f rustfs-test-mqtt >/dev/null 2>&1 || true
+          cat <<'EOF' | sudo tee /tmp/rustfs-mosquitto.conf >/dev/null
+          listener 1883 0.0.0.0
+          allow_anonymous true
+          EOF
+          sudo docker run -d --name rustfs-test-mqtt -p 1883:1883 \
+            -v /tmp/rustfs-mosquitto.conf:/mosquitto/config/mosquitto.conf:ro \
+            eclipse-mosquitto:2 >/dev/null
+          for _ in {1..10}; do
+            if ss -tln 2>/dev/null | grep -q ':1883'; then
+              break
+            fi
+            sleep 1
+          done
+          ss -tln 2>/dev/null | grep -q ':1883' || {
+            echo 'mosquitto container is not listening on 1883'
+            sudo docker logs rustfs-test-mqtt || true
+            exit 1
+          }
+
+      - name: Run tier suite
+        id: test
+        continue-on-error: true
+        env:
+          PACKAGE_URL_INPUT: ${{ inputs.package_url }}
+          RUSTFS_VERSION_INPUT: ${{ inputs.rustfs_version }}
+        run: |
+          set -euo pipefail
+          LOG_FILE="${TIER_ARTIFACTS_DIR}/rustfs-tier.log"
+          chmod +x auto-testing/rustfs-tier-test.sh
+          RC_BIN="$(command -v rc)"
+          PACKAGE_URL="${PACKAGE_URL_INPUT}"
+          RUSTFS_VERSION="${RUSTFS_VERSION_INPUT}"
+          ARGS=(
+            --all-topologies
+            -y
+            --log-file "${LOG_FILE}"
+            --rc-bin "${RC_BIN}"
+            --artifacts-dir "${TIER_ARTIFACTS_DIR}"
+          )
+          if [ -n "${RUSTFS_EXPECTED_RC_SHA256}" ]; then
+            ARGS+=(--expected-rc-sha256 "${RUSTFS_EXPECTED_RC_SHA256}")
+          fi
+          if [ -n "${PACKAGE_URL}" ]; then
+            ARGS+=(--package-url "${PACKAGE_URL}")
+          elif [ -n "${RUSTFS_VERSION}" ]; then
+            ARGS+=(--version "${RUSTFS_VERSION}")
+          else
+            ARGS+=(--package-url "${RUSTFS_NIGHTLY_PACKAGE_URL}")
+          fi
+          ./auto-testing/rustfs-tier-test.sh "${ARGS[@]}"
+
+      - name: Inject diagnostic case failure
+        if: ${{ always() && steps.evidence.outcome == 'success' && inputs.force_case_failure }}
+        run: |
+          set -euo pipefail
+          RESULT_FILE="${TIER_ARTIFACTS_DIR}/cases/single-single--TIER-101.json"
+          test -s "${RESULT_FILE}"
+          TMP_FILE="$(mktemp "${TIER_ARTIFACTS_DIR}/cases/.forced.XXXXXX")"
+          jq '.status = "FAIL" | .case_rc = 97' "${RESULT_FILE}" > "${TMP_FILE}"
+          mv "${TMP_FILE}" "${RESULT_FILE}"
+
+      - name: Generate report
+        if: ${{ always() && steps.evidence.outcome == 'success' }}
+        env:
+          PACKAGE_URL_INPUT: ${{ inputs.package_url }}
+          RUSTFS_VERSION_INPUT: ${{ inputs.rustfs_version }}
+          TEST_OUTCOME: ${{ steps.test.outcome }}
+          RUN_URL: ${{ github.server_url }}/${{ github.repository }}/actions/runs/${{ github.run_id }}
+          TRIGGER_NAME: ${{ github.event_name }}
+        run: |
+          set -euo pipefail
+          test -d "${TIER_ARTIFACTS_DIR}"
+          test ! -L "${TIER_ARTIFACTS_DIR}"
+          LOG_FILE="${TIER_ARTIFACTS_DIR}/rustfs-tier.log"
+          REPORT_FILE="${TIER_ARTIFACTS_DIR}/rustfs-tier-report.md"
+          CASE_TABLE="${TIER_ARTIFACTS_DIR}/rustfs-tier-cases.md"
+          GATE_RC_FILE="${TIER_ARTIFACTS_DIR}/rustfs-tier-gate.rc"
+          PACKAGE_URL="${PACKAGE_URL_INPUT}"
+          RUSTFS_VERSION="${RUSTFS_VERSION_INPUT}"
+          if [ -n "${PACKAGE_URL}" ]; then
+            PACKAGE_SOURCE="${PACKAGE_URL}"
+          elif [ -n "${RUSTFS_VERSION}" ]; then
+            PACKAGE_SOURCE="version ${RUSTFS_VERSION}"
+          else
+            PACKAGE_SOURCE="${RUSTFS_NIGHTLY_PACKAGE_URL}"
+          fi
+          set +e
+          python3 auto-testing/rustfs_tier_report.py \
+            --results-dir "${TIER_ARTIFACTS_DIR}/cases" \
+            --provenance "${TIER_ARTIFACTS_DIR}/provenance.json" \
+            --output "${CASE_TABLE}"
+          CASE_GATE_RC=$?
+          set -e
+          printf '%s\n' "${CASE_GATE_RC}" > "${GATE_RC_FILE}"
+          if [ ! -s "${CASE_TABLE}" ]; then
+            {
+              echo "## Case Summary"
+              echo ""
+              echo "Structured report generation failed before producing output (exit ${CASE_GATE_RC})."
+            } > "${CASE_TABLE}"
+          fi
+          {
+            echo "# RustFS tier test report"
+            echo ""
+            echo "- Run: ${RUN_URL}"
+            echo "- Trigger: ${TRIGGER_NAME}"
+            echo "- Package: ${PACKAGE_SOURCE}"
+            echo "- Test Step Outcome: ${TEST_OUTCOME}"
+            echo "- Structured Gate Exit: ${CASE_GATE_RC}"
+            echo ""
+            cat "${CASE_TABLE}"
+            echo ""
+            echo "## Log tail"
+            echo '```text'
+            tail -n 200 "${LOG_FILE}" || true
+            echo '```'
+          } | tee "${REPORT_FILE}"
+          cat "${REPORT_FILE}" >> "${GITHUB_STEP_SUMMARY}"
+
+      - name: Upload functional report to dashboard
+        if: ${{ always() && steps.evidence.outcome == 'success' }}
+        continue-on-error: true
+        env:
+          GH_TOKEN: ${{ env.PF_TESTING_GH_TOKEN }}
+          REPORT_FILE: ${{ env.TIER_ARTIFACTS_DIR }}/rustfs-tier-report.md
+          SUITE: tier
+        run: |
+          set -euo pipefail
+          if [ -z "${GH_TOKEN:-}" ]; then
+            echo "PF_TESTING_GH_TOKEN is not configured; skipping dashboard upload"
+            exit 0
+          fi
+          DATE="$(date -u +%Y-%m-%d)"
+          REPORT_PATH="functional-reports/${SUITE}/${DATE}.md"
+          CONTENT="$(python3 -c 'import base64,sys;print(base64.b64encode(open(sys.argv[1],"rb").read()).decode())' "${REPORT_FILE}")"
+          SHA="$(gh api "repos/rustfs/dashboard/contents/${REPORT_PATH}" -q '.sha' 2>/dev/null || true)"
+          if [ -n "${SHA}" ]; then
+            jq -n --arg msg "report(${SUITE}): ${DATE}" --arg content "${CONTENT}" --arg sha "${SHA}" \
+              '{message:$msg, content:$content, sha:$sha}' \
+              | gh api --method PUT "repos/rustfs/dashboard/contents/${REPORT_PATH}" --input - >/dev/null
+          else
+            jq -n --arg msg "report(${SUITE}): ${DATE}" --arg content "${CONTENT}" \
+              '{message:$msg, content:$content}' \
+              | gh api --method PUT "repos/rustfs/dashboard/contents/${REPORT_PATH}" --input - >/dev/null
+          fi
+
+      - name: Verify required tier evidence
+        id: evidence_verify
+        if: ${{ always() && steps.evidence.outcome == 'success' }}
+        run: |
+          set -euo pipefail
+          failed=0
+          for name in \
+            rustfs-tier.log \
+            rustfs-tier-report.md \
+            rustfs-tier-cases.md \
+            rustfs-tier-gate.rc \
+            provenance.json; do
+            if [ ! -s "${TIER_ARTIFACTS_DIR}/${name}" ]; then
+              echo "required tier evidence is missing or empty: ${name}" >&2
+              failed=1
+            fi
+          done
+          for name in cases logs; do
+            if [ ! -d "${TIER_ARTIFACTS_DIR}/${name}" ]; then
+              echo "required tier evidence directory is missing: ${name}" >&2
+              failed=1
+            fi
+          done
+          if ! find "${TIER_ARTIFACTS_DIR}/cases" -maxdepth 1 -type f -name '*.json' -print -quit 2>/dev/null | grep -q .; then
+            echo "no atomic tier case result was produced" >&2
+            failed=1
+          fi
+          [ "${failed}" -eq 0 ]
+
+      - name: Upload report and logs
+        if: ${{ always() && steps.evidence.outcome == 'success' }}
+        uses: actions/upload-artifact@b7c566a772e6b6bfb58ed0dc250532a479d7789f # v6
+        with:
+          name: rustfs-tier-test-${{ github.run_id }}-${{ github.run_attempt }}
+          path: ${{ env.TIER_ARTIFACTS_DIR }}/
+          if-no-files-found: error
+
+      - name: Cleanup environment (after)
+        if: always()
+        run: |
+          set -euo pipefail
+          sudo docker rm -f rustfs-test-mqtt >/dev/null 2>&1 || true
+          sudo rm -f /tmp/rustfs-mosquitto.conf
+          read -r -a NODES <<< "${RUSTFS_NODES:-vm000 vm001 vm002}"
+          SSH_USER="${RUSTFS_SSH_USER:-azureuser}"
+          for node in "${NODES[@]}"; do
+            ssh -o BatchMode=yes -o ConnectTimeout=10 -o StrictHostKeyChecking=accept-new "${SSH_USER}@${node}" '
+              set -euo pipefail
+              SUDO=""; [ "$(id -u)" -ne 0 ] && SUDO="sudo -n"
+              ${SUDO} systemctl stop rustfs 2>/dev/null || true
+              if ${SUDO} dpkg -l rustfs 2>/dev/null | grep -q "^ii"; then
+                ${SUDO} dpkg -P rustfs
+              fi
+              for i in 1 2 3 4; do ${SUDO} rm -rf /data/rustfs${i}/mnmd; done
+              ${SUDO} rm -rf /var/log/rustfs /var/lib/rustfs/kms
+            '
+          done
+
+      - name: Enforce tier suite result
+        id: gate
+        if: always()
+        env:
+          EVIDENCE_OUTCOME: ${{ steps.evidence.outcome }}
+          TEST_OUTCOME: ${{ steps.test.outcome }}
+          GATE_RC_FILE: ${{ env.TIER_ARTIFACTS_DIR }}/rustfs-tier-gate.rc
+        run: |
+          set -euo pipefail
+          failed=0
+          if [ "${EVIDENCE_OUTCOME}" != "success" ]; then
+            echo "tier evidence directory initialization is ${EVIDENCE_OUTCOME}, expected success" >&2
+            failed=1
+          fi
+          if [ "${TEST_OUTCOME}" != "success" ]; then
+            echo "tier suite step outcome is ${TEST_OUTCOME}, expected success" >&2
+            failed=1
+          fi
+          if [ "${EVIDENCE_OUTCOME}" != "success" ]; then
+            echo "structured gate result is unavailable because evidence initialization failed" >&2
+          elif [ ! -s "${GATE_RC_FILE}" ]; then
+            echo "structured gate result is missing" >&2
+            failed=1
+          else
+            GATE_RC="$(tr -d '[:space:]' < "${GATE_RC_FILE}")"
+            if ! [[ "${GATE_RC}" =~ ^[0-9]+$ ]] || [ "${GATE_RC}" -ne 0 ]; then
+              echo "structured 56-case gate failed with exit ${GATE_RC:-invalid}" >&2
+              failed=1
+            fi
+          fi
+          [ "${failed}" -eq 0 ]
+
+      - name: File failure issue in rustfs/backlog
+        if: ${{ always() && (failure() || steps.test.outcome == 'failure' || steps.test.outcome == 'cancelled' || steps.evidence_verify.outcome == 'failure' || steps.evidence_verify.outcome == 'cancelled' || steps.gate.outcome == 'failure' || steps.gate.outcome == 'cancelled') }}
+        continue-on-error: true
+        env:
+          GH_TOKEN: ${{ secrets.PF_TESTING_GH_TOKEN }}
+          SUITE: 'tier'
+          SUITE_LABEL: 'Tier'
+          RUN_URL: ${{ github.server_url }}/${{ github.repository }}/actions/runs/${{ github.run_id }}
+          EVIDENCE_DIR: ${{ env.TIER_ARTIFACTS_DIR }}
+          EVIDENCE_OUTCOME: ${{ steps.evidence.outcome }}
+          VERIFY_OUTCOME: ${{ steps.evidence_verify.outcome }}
+          GATE_OUTCOME: ${{ steps.gate.outcome }}
+          REPORT_FILE: ${{ env.TIER_ARTIFACTS_DIR }}/rustfs-tier-report.md
+          LOG_FILE: ${{ env.TIER_ARTIFACTS_DIR }}/rustfs-tier.log
+        run: |
+          set -euo pipefail
+          if [ -z "${GH_TOKEN:-}" ]; then
+            echo "PF_TESTING_GH_TOKEN is not configured; skipping backlog issue"
+            exit 0
+          fi
+          TITLE="[functional][${SUITE}] ${SUITE_LABEL} suite failed (run ${GITHUB_RUN_ID})"
+          EXISTING="$(gh issue list -R rustfs/backlog --state all \
+            --search "in:title \"run ${GITHUB_RUN_ID}\"" \
+            --json number --jq '.[].number' || true)"
+          if [ -n "${EXISTING}" ]; then
+            echo "backlog issue already exists for run ${GITHUB_RUN_ID}; skipping"
+            exit 0
+          fi
+          redact() {
+            sed -E \
+              -e 's/(RUSTFS_(ACCESS_KEY|SECRET_KEY)[=: ]+)[^[:space:]]+/\1[REDACTED]/Ig' \
+              -e 's/(Authorization:).*/\1 [REDACTED]/Ig' \
+              -e 's/(X-Amz-Signature=)[^&[:space:]]+/\1[REDACTED]/Ig' \
+              -e 's/^.*(password|secret|token)[=: ].*/[REDACTED SENSITIVE LINE]/Ig'
+          }
+          BODY_FILE="$(mktemp)"
+          {
+            echo "The **${SUITE_LABEL}** functional suite failed."
+            echo ""
+            echo "- Suite: \`${SUITE}\`"
+            echo "- Run: ${RUN_URL}"
+            echo "- Trigger: ${GITHUB_EVENT_NAME}"
+            echo "- Date: $(date -u +%Y-%m-%d)"
+            echo "- Evidence initialization: ${EVIDENCE_OUTCOME}"
+            echo "- Evidence verification: ${VERIFY_OUTCOME}"
+            echo "- Final gate: ${GATE_OUTCOME}"
+            echo ""
+            echo "## Report (errors and symptoms)"
+            echo ""
+            if [ "${EVIDENCE_OUTCOME}" != "success" ]; then
+              echo "(the run evidence directory was rejected; its contents were not read)"
+            elif [ ! -d "${EVIDENCE_DIR}" ] || [ -L "${EVIDENCE_DIR}" ]; then
+              echo "(the run evidence directory is missing or unsafe; its contents were not read)"
+            elif [ -s "${REPORT_FILE}" ]; then
+              redact < "${REPORT_FILE}"
+            elif [ -s "${LOG_FILE:-}" ]; then
+              echo "(report file missing; log tail below)"
+              echo ""
+              tail -n 200 "${LOG_FILE}" | redact
+            else
+              echo "(no report or log file was produced)"
+            fi
+          } | head -c 55000 > "${BODY_FILE}"
+          gh label create functional-test -R rustfs/backlog --color d73a4a 2>/dev/null || true
+          if ! gh issue create -R rustfs/backlog --title "${TITLE}" \
+              --body-file "${BODY_FILE}" --label functional-test; then
+            gh issue create -R rustfs/backlog --title "${TITLE}" --body-file "${BODY_FILE}"
+          fi
+          echo "filed backlog issue for suite ${SUITE}"
+
+      - name: "Continue functional chain (next: Storage engine)"
+        # Only chain-triggered runs forward to the next suite; standalone
+        # workflow_dispatch runs stop after their own cleanup. A failed
+        # handoff must never pass silently: it retries, then files an alert
+        # issue in rustfs/backlog so a stalled chain is visible.
+        if: ${{ always() && github.event_name == 'repository_dispatch' }}
+        continue-on-error: true
+        env:
+          GH_TOKEN: ${{ secrets.PF_TESTING_GH_TOKEN }}
+        run: |
+          set -uo pipefail
+          if [ -z "${GH_TOKEN:-}" ]; then
+            echo "PF_TESTING_GH_TOKEN is not configured; cannot dispatch the next suite" >&2
+            exit 1
+          fi
+          DISPATCHED=0
+          for attempt in 1 2 3; do
+            if gh api --method POST repos/rustfs/rustfs/dispatches \
+              -f event_type='rustfs-chain-storage' \
+              -F 'client_payload[from_suite]=tier'; then
+              echo "dispatched next suite Storage engine (attempt ${attempt})"
+              DISPATCHED=1
+              break
+            fi
+            echo "dispatch attempt ${attempt} failed; retrying in ${attempt}0s" >&2
+            sleep "${attempt}0"
+          done
+          if [ "${DISPATCHED:-0}" -ne 1 ]; then
+            echo "ERROR: functional chain stalled: could not dispatch Storage engine after 3 attempts" >&2
+            TITLE="[functional][chain] stalled after tier (run ${GITHUB_RUN_ID})"
+            BODY_FILE="$(mktemp)"
+            {
+              echo "The functional chain could not hand off from **tier** to **Storage engine** after 3 attempts."
+              echo ""
+              echo "- Failed suite job: ${GITHUB_SERVER_URL}/${GITHUB_REPOSITORY}/actions/runs/${GITHUB_RUN_ID}"
+              echo "- Expected next event: 'rustfs-chain-storage'"
+              echo "- Likely cause: PF_TESTING_GH_TOKEN lacks contents:write on rustfs/rustfs, or the GitHub API was unavailable."
+              echo "- Recovery: re-dispatch manually with"
+              FENCE="$(printf "\x60\x60\x60")"; echo "  ${FENCE}"
+              echo "  gh api --method POST repos/rustfs/rustfs/dispatches -f event_type='rustfs-chain-storage'"
+              FENCE="$(printf "\x60\x60\x60")"; echo "  ${FENCE}"
+            } > "${BODY_FILE}"
+            gh issue create -R rustfs/backlog --title "${TITLE}" \
+              --body-file "${BODY_FILE}" --label functional-test \
+              || gh issue create -R rustfs/backlog --title "${TITLE}" --body-file "${BODY_FILE}" \
+              || echo "could not file the stall alert issue either; check the token" >&2
+            exit 1
+          fi
+
+      - name: Notify on failure
+        if: failure()
+        run: |
+          echo "RustFS tier suite failed"
+          echo "See the uploaded report and log artifacts for details."
@@ -0,0 +1,444 @@
+# Copyright 2024 RustFS Team
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#     http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+
+name: RustFS Upgrade Test
+
+on:
+  workflow_dispatch:
+    inputs:
+      from_version:
+        description: 'OLD RustFS release tag (e.g. 1.0.0-rc.4-preview.1)'
+        required: false
+        default: '1.0.0-rc.4-preview.1'
+      from_url:
+        description: 'OLD .deb URL. Overrides from_version.'
+        required: false
+        type: string
+      to_version:
+        description: 'NEW RustFS release tag (leave empty for latest nightly)'
+        required: false
+      to_url:
+        description: 'NEW .deb URL. Overrides to_version / nightly default.'
+        required: false
+        type: string
+      topology:
+        description: 'Topology to run (all = SNSD, SNMD, MNMD)'
+        type: choice
+        options:
+          - all
+          - single-single
+          - single-multi
+          - multi-multi
+        default: all
+      backends:
+        description: 'KMS backends to run (local,vault-kv2)'
+        required: false
+        default: 'local,vault-kv2'
+      cleanup_before:
+        description: 'Reset the nodes before the test (DESTROYS existing data/config)'
+        type: boolean
+        default: true
+      cleanup_after:
+        description: 'Reset the nodes after the test (DESTROYS test data/config)'
+        type: boolean
+        default: true
+  repository_dispatch:
+    # Functional-chain entry: dispatched by rustfs-functional-chain.yml.
+    types: [rustfs-chain-upgrade]
+
+permissions:
+  contents: read
+
+concurrency:
+  group: rustfs-shared-functional-tests
+  cancel-in-progress: false
+
+defaults:
+  run:
+    shell: bash
+
+env:
+  RUSTFS_ACCESS_KEY: ${{ secrets.RUSTFS_ACCESS_KEY }}
+  RUSTFS_SECRET_KEY: ${{ secrets.RUSTFS_SECRET_KEY }}
+  RUSTFS_NODES: ${{ secrets.RUSTFS_NODES || vars.RUSTFS_NODES }}
+  RUSTFS_SSH_USER: ${{ secrets.RUSTFS_SSH_USER || vars.RUSTFS_SSH_USER }}
+  RUSTFS_NIGHTLY_PACKAGE_URL: ${{ vars.RUSTFS_NIGHTLY_PACKAGE_URL || 'https://dl.rustfs.com/artifacts/rustfs/packages/nightly/rustfs-nightly-latest.deb' }}
+  PF_TESTING_GH_TOKEN: ${{ secrets.PF_TESTING_GH_TOKEN }}
+
+jobs:
+  upgrade-test:
+    runs-on: smoke-testing
+    continue-on-error: true
+    timeout-minutes: 420
+    if: ${{ github.event_name == 'workflow_dispatch' || github.event_name == 'repository_dispatch' }}
+    steps:
+      # auto-testing is private: clone it with the dedicated PF token (not
+      # GITHUB_TOKEN) and retry transient GitHub/network failures.
+      - name: Checkout auto-testing scripts (with retry)
+        env:
+          GH_TOKEN: ${{ secrets.PF_TESTING_GH_TOKEN }}
+        run: |
+          set -euo pipefail
+          rm -rf auto-testing
+          for attempt in 1 2 3 4 5; do
+            if gh repo clone rustfs/auto-testing auto-testing -- --depth 1 --quiet; then
+              echo "auto-testing cloned (attempt ${attempt})"
+              exit 0
+            fi
+            rm -rf auto-testing
+            echo "clone attempt ${attempt} failed; retrying in $((attempt * 15))s" >&2
+            sleep $((attempt * 15))
+          done
+          echo "ERROR: unable to clone rustfs/auto-testing after 5 attempts" >&2
+          exit 1
+
+      - name: Show environment
+        run: |
+          uname -a
+          jq --version
+          openssl version
+          aws --version || true
+          docker --version || true
+          df -h /data | tail -1
+
+      - name: Cleanup environment (before)
+        if: ${{ inputs.cleanup_before != 'false' || github.event_name != 'workflow_dispatch' }}
+        run: |
+          set -euo pipefail
+          read -r -a NODES <<< "${RUSTFS_NODES:-vm000 vm001 vm002}"
+          SSH_USER="${RUSTFS_SSH_USER:-azureuser}"
+          for node in "${NODES[@]}"; do
+            ssh -o BatchMode=yes -o ConnectTimeout=10 -o StrictHostKeyChecking=accept-new "${SSH_USER}@${node}" '
+              set -euo pipefail
+              SUDO=""; [ "$(id -u)" -ne 0 ] && SUDO="sudo -n"
+              ${SUDO} systemctl stop rustfs 2>/dev/null || true
+              if ${SUDO} dpkg -l rustfs 2>/dev/null | grep -q "^ii"; then
+                ${SUDO} dpkg -P rustfs
+              fi
+              for i in 1 2 3 4; do ${SUDO} rm -rf /data/rustfs${i}/mnmd; done
+              ${SUDO} rm -rf /var/log/rustfs /var/lib/rustfs/kms /var/lib/rustfs/kms-backup
+            '
+          done
+
+      - name: Ensure docker (Vault container)
+        run: |
+          if ! command -v docker >/dev/null 2>&1; then
+            sudo apt-get update
+            sudo apt-get install -y docker.io
+          fi
+          sudo systemctl enable --now docker
+          docker info >/dev/null 2>&1 || sudo docker info >/dev/null 2>&1
+
+      - name: Run upgrade compatibility suite
+        id: test
+        continue-on-error: true
+        env:
+          LOG_FILE: /tmp/rustfs-upgrade.log
+        run: |
+          set -euo pipefail
+          chmod +x auto-testing/rustfs-upgrade-test.sh
+          FROM_URL='${{ inputs.from_url }}'
+          FROM_VERSION='${{ inputs.from_version }}'
+          TO_URL='${{ inputs.to_url }}'
+          TO_VERSION='${{ inputs.to_version }}'
+          TOPOLOGY='${{ inputs.topology }}'
+          BACKENDS='${{ inputs.backends }}'
+          ARGS=(-y --log-file "${LOG_FILE}")
+          if [ "${TOPOLOGY}" = "all" ] || [ -z "${TOPOLOGY}" ] || [ "${TOPOLOGY}" = "null" ]; then
+            ARGS+=(--all-topologies)
+          else
+            ARGS+=(--topology "${TOPOLOGY}")
+          fi
+          if [ -n "${BACKENDS}" ] && [ "${BACKENDS}" != "null" ]; then
+            ARGS+=(--backends "${BACKENDS}")
+          fi
+          if [ -n "${FROM_URL}" ]; then
+            ARGS+=(--from-url "${FROM_URL}")
+          elif [ -n "${FROM_VERSION}" ] && [ "${FROM_VERSION}" != "null" ]; then
+            ARGS+=(--from-version "${FROM_VERSION}")
+          fi
+          if [ -n "${TO_URL}" ]; then
+            ARGS+=(--to-url "${TO_URL}")
+          elif [ -n "${TO_VERSION}" ] && [ "${TO_VERSION}" != "null" ]; then
+            ARGS+=(--to-version "${TO_VERSION}")
+          else
+            ARGS+=(--to-url "${RUSTFS_NIGHTLY_PACKAGE_URL}")
+          fi
+          ./auto-testing/rustfs-upgrade-test.sh "${ARGS[@]}"
+
+      - name: Generate report
+        if: always()
+        env:
+          LOG_FILE: /tmp/rustfs-upgrade.log
+          REPORT_FILE: /tmp/rustfs-upgrade-report.md
+        run: |
+          set -euo pipefail
+          FROM_URL='${{ inputs.from_url }}'
+          FROM_VERSION='${{ inputs.from_version }}'
+          TO_URL='${{ inputs.to_url }}'
+          TO_VERSION='${{ inputs.to_version }}'
+          if [ -n "${FROM_URL}" ]; then
+            FROM_SOURCE="${FROM_URL}"
+          elif [ -n "${FROM_VERSION}" ]; then
+            FROM_SOURCE="version ${FROM_VERSION}"
+          else
+            FROM_SOURCE="release (default)"
+          fi
+          if [ -n "${TO_URL}" ]; then
+            TO_SOURCE="${TO_URL}"
+          elif [ -n "${TO_VERSION}" ]; then
+            TO_SOURCE="version ${TO_VERSION}"
+          else
+            TO_SOURCE="${RUSTFS_NIGHTLY_PACKAGE_URL}"
+          fi
+          CASE_TABLE="/tmp/rustfs-upgrade-cases.md"
+          python3 - "${LOG_FILE}" "${CASE_TABLE}" <<'PY'
+          import re
+          import sys
+
+          log_file, out_file = sys.argv[1], sys.argv[2]
+          ansi = re.compile(r'\x1b\[[0-9;]*m')
+          start_re = re.compile(r'^---\s+([A-Z]+-[0-9]+)\s+(.+?)\s+---$')
+          done_re = re.compile(r'^\[(PASS|FAIL|UNSUPPORTED)\]\s+([A-Z]+-[0-9]+)\b')
+
+          rows = []
+          index = {}
+          try:
+            with open(log_file, 'r', encoding='utf-8', errors='replace') as fh:
+              for raw in fh:
+                line = ansi.sub('', raw).strip()
+                m = start_re.match(line)
+                if m:
+                  case_id, name = m.group(1), m.group(2)
+                  if case_id not in index:
+                    index[case_id] = len(rows)
+                    rows.append([case_id, name, 'RUNNING'])
+                  continue
+                m = done_re.match(line)
+                if m:
+                  status, case_id = m.group(1), m.group(2)
+                  if case_id in index:
+                    rows[index[case_id]][2] = status
+                  else:
+                    rows.append([case_id, case_id, status])
+                    index[case_id] = len(rows) - 1
+          except FileNotFoundError:
+            rows = []
+
+          counts = {'PASS': 0, 'FAIL': 0, 'UNSUPPORTED': 0, 'RUNNING': 0}
+          for _, _, status in rows:
+            counts[status] = counts.get(status, 0) + 1
+
+          with open(out_file, 'w', encoding='utf-8') as out:
+            out.write('## Case Summary\n\n')
+            out.write(f"- Total: {len(rows)}\\n")
+            out.write(f"- PASS: {counts.get('PASS', 0)}\\n")
+            out.write(f"- FAIL: {counts.get('FAIL', 0)}\\n")
+            out.write(f"- UNSUPPORTED: {counts.get('UNSUPPORTED', 0)}\\n")
+            out.write('\\n')
+            out.write('| Case | Name | Status |\\n')
+            out.write('| --- | --- | --- |\\n')
+            for case_id, name, status in rows:
+              out.write(f'| {case_id} | {name} | {status} |\\n')
+          PY
+          {
+            echo "# RustFS upgrade compatibility report"
+            echo ""
+            echo "- Run: ${{ github.server_url }}/${{ github.repository }}/actions/runs/${{ github.run_id }}"
+            echo "- Trigger: ${{ github.event_name }}"
+            echo "- From: ${FROM_SOURCE}"
+            echo "- To: ${TO_SOURCE}"
+            echo "- Test Step Outcome: ${{ steps.test.outcome }}"
+            echo ""
+            cat "${CASE_TABLE}" || true
+            echo ""
+            echo "## Log tail"
+            echo '```text'
+            tail -n 200 "${LOG_FILE}" || true
+            echo '```'
+          } | tee "${REPORT_FILE}"
+          cat "${REPORT_FILE}" >> "${GITHUB_STEP_SUMMARY}"
+
+      - name: Upload functional report to dashboard
+        if: always()
+        continue-on-error: true
+        env:
+          GH_TOKEN: ${{ env.PF_TESTING_GH_TOKEN }}
+          REPORT_FILE: /tmp/rustfs-upgrade-report.md
+          SUITE: upgrade
+        run: |
+          set -euo pipefail
+          if [ -z "${GH_TOKEN:-}" ]; then
+            echo "PF_TESTING_GH_TOKEN is not configured; skipping dashboard upload"
+            exit 0
+          fi
+          DATE="$(date -u +%Y-%m-%d)"
+          REPORT_PATH="functional-reports/${SUITE}/${DATE}.md"
+          CONTENT="$(python3 -c 'import base64,sys;print(base64.b64encode(open(sys.argv[1],"rb").read()).decode())' "${REPORT_FILE}")"
+          SHA="$(gh api "repos/rustfs/dashboard/contents/${REPORT_PATH}" -q '.sha' 2>/dev/null || true)"
+          if [ -n "${SHA}" ]; then
+            jq -n --arg msg "report(${SUITE}): ${DATE}" --arg content "${CONTENT}" --arg sha "${SHA}" \
+              '{message:$msg, content:$content, sha:$sha}' \
+              | gh api --method PUT "repos/rustfs/dashboard/contents/${REPORT_PATH}" --input - >/dev/null
+          else
+            jq -n --arg msg "report(${SUITE}): ${DATE}" --arg content "${CONTENT}" \
+              '{message:$msg, content:$content}' \
+              | gh api --method PUT "repos/rustfs/dashboard/contents/${REPORT_PATH}" --input - >/dev/null
+          fi
+
+      - name: File failure issue in rustfs/backlog
+        if: ${{ always() && (failure() || steps.test.outcome == 'failure' || steps.test.outcome == 'cancelled') }}
+        continue-on-error: true
+        env:
+          GH_TOKEN: ${{ secrets.PF_TESTING_GH_TOKEN }}
+          SUITE: 'upgrade'
+          SUITE_LABEL: 'Upgrade compatibility'
+          RUN_URL: ${{ github.server_url }}/${{ github.repository }}/actions/runs/${{ github.run_id }}
+          REPORT_FILE: '/tmp/rustfs-upgrade-report.md'
+          LOG_FILE: '/tmp/rustfs-upgrade.log'
+        run: |
+          set -euo pipefail
+          if [ -z "${GH_TOKEN:-}" ]; then
+            echo "PF_TESTING_GH_TOKEN is not configured; skipping backlog issue"
+            exit 0
+          fi
+          TITLE="[functional][${SUITE}] ${SUITE_LABEL} suite failed (run ${GITHUB_RUN_ID})"
+          EXISTING="$(gh issue list -R rustfs/backlog --state all \
+            --search "in:title \"run ${GITHUB_RUN_ID}\"" \
+            --json number --jq '.[].number' || true)"
+          if [ -n "${EXISTING}" ]; then
+            echo "backlog issue already exists for run ${GITHUB_RUN_ID}; skipping"
+            exit 0
+          fi
+          redact() {
+            sed -E \
+              -e 's/(RUSTFS_(ACCESS_KEY|SECRET_KEY)[=: ]+)[^[:space:]]+/\1[REDACTED]/Ig' \
+              -e 's/(Authorization:).*/\1 [REDACTED]/Ig' \
+              -e 's/(X-Amz-Signature=)[^&[:space:]]+/\1[REDACTED]/Ig' \
+              -e 's/^.*(password|secret|token)[=: ].*/[REDACTED SENSITIVE LINE]/Ig'
+          }
+          BODY_FILE="$(mktemp)"
+          {
+            echo "The **${SUITE_LABEL}** functional suite failed."
+            echo ""
+            echo "- Suite: \`${SUITE}\`"
+            echo "- Run: ${RUN_URL}"
+            echo "- Trigger: ${GITHUB_EVENT_NAME}"
+            echo "- Date: $(date -u +%Y-%m-%d)"
+            echo ""
+            echo "## Report (errors and symptoms)"
+            echo ""
+            if [ -s "${REPORT_FILE}" ]; then
+              redact < "${REPORT_FILE}"
+            elif [ -s "${LOG_FILE:-}" ]; then
+              echo "(report file missing; log tail below)"
+              echo ""
+              tail -n 200 "${LOG_FILE}" | redact
+            else
+              echo "(no report or log file was produced)"
+            fi
+          } | head -c 55000 > "${BODY_FILE}"
+          gh label create functional-test -R rustfs/backlog --color d73a4a 2>/dev/null || true
+          if ! gh issue create -R rustfs/backlog --title "${TITLE}" \
+              --body-file "${BODY_FILE}" --label functional-test; then
+            gh issue create -R rustfs/backlog --title "${TITLE}" --body-file "${BODY_FILE}"
+          fi
+          echo "filed backlog issue for suite ${SUITE}"
+
+      - name: Upload report and logs
+        if: always()
+        uses: actions/upload-artifact@b7c566a772e6b6bfb58ed0dc250532a479d7789f # v6
+        with:
+          name: rustfs-upgrade-test-${{ github.run_id }}
+          path: |
+            /tmp/rustfs-upgrade-report.md
+            /tmp/rustfs-upgrade.*/*
+          if-no-files-found: ignore
+          retention-days: 3
+
+      - name: Cleanup environment (after)
+        if: ${{ always() && (inputs.cleanup_after != 'false' || github.event_name != 'workflow_dispatch') }}
+        run: |
+          set -euo pipefail
+          read -r -a NODES <<< "${RUSTFS_NODES:-vm000 vm001 vm002}"
+          SSH_USER="${RUSTFS_SSH_USER:-azureuser}"
+          for node in "${NODES[@]}"; do
+            ssh -o BatchMode=yes -o ConnectTimeout=10 -o StrictHostKeyChecking=accept-new "${SSH_USER}@${node}" '
+              set -euo pipefail
+              SUDO=""; [ "$(id -u)" -ne 0 ] && SUDO="sudo -n"
+              ${SUDO} systemctl stop rustfs 2>/dev/null || true
+              if ${SUDO} dpkg -l rustfs 2>/dev/null | grep -q "^ii"; then
+                ${SUDO} dpkg -P rustfs
+              fi
+              for i in 1 2 3 4; do ${SUDO} rm -rf /data/rustfs${i}/mnmd; done
+              ${SUDO} rm -rf /var/log/rustfs /var/lib/rustfs/kms /var/lib/rustfs/kms-backup
+            '
+          done
+
+      - name: "Continue functional chain (next: S3 compatibility)"
+        # Only chain-triggered runs forward to the next suite; standalone
+        # workflow_dispatch runs stop after their own cleanup. A failed
+        # handoff must never pass silently: it retries, then files an alert
+        # issue in rustfs/backlog so a stalled chain is visible.
+        if: ${{ always() && github.event_name == 'repository_dispatch' }}
+        continue-on-error: true
+        env:
+          GH_TOKEN: ${{ secrets.PF_TESTING_GH_TOKEN }}
+        run: |
+          set -uo pipefail
+          if [ -z "${GH_TOKEN:-}" ]; then
+            echo "PF_TESTING_GH_TOKEN is not configured; cannot dispatch the next suite" >&2
+            exit 1
+          fi
+          DISPATCHED=0
+          for attempt in 1 2 3; do
+            if gh api --method POST repos/rustfs/rustfs/dispatches \
+              -f event_type='rustfs-chain-s3' \
+              -F 'client_payload[from_suite]=upgrade'; then
+              echo "dispatched next suite S3 compatibility (attempt ${attempt})"
+              DISPATCHED=1
+              break
+            fi
+            echo "dispatch attempt ${attempt} failed; retrying in ${attempt}0s" >&2
+            sleep "${attempt}0"
+          done
+          if [ "${DISPATCHED:-0}" -ne 1 ]; then
+            echo "ERROR: functional chain stalled: could not dispatch S3 compatibility after 3 attempts" >&2
+            TITLE="[functional][chain] stalled after upgrade (run ${GITHUB_RUN_ID})"
+            BODY_FILE="$(mktemp)"
+            {
+              echo "The functional chain could not hand off from **upgrade** to **S3 compatibility** after 3 attempts."
+              echo ""
+              echo "- Failed suite job: ${GITHUB_SERVER_URL}/${GITHUB_REPOSITORY}/actions/runs/${GITHUB_RUN_ID}"
+              echo "- Expected next event: 'rustfs-chain-s3'"
+              echo "- Likely cause: PF_TESTING_GH_TOKEN lacks contents:write on rustfs/rustfs, or the GitHub API was unavailable."
+              echo "- Recovery: re-dispatch manually with"
+              FENCE="$(printf "\x60\x60\x60")"; echo "  ${FENCE}"
+              echo "  gh api --method POST repos/rustfs/rustfs/dispatches -f event_type='rustfs-chain-s3'"
+              FENCE="$(printf "\x60\x60\x60")"; echo "  ${FENCE}"
+            } > "${BODY_FILE}"
+            gh issue create -R rustfs/backlog --title "${TITLE}" \
+              --body-file "${BODY_FILE}" --label functional-test \
+              || gh issue create -R rustfs/backlog --title "${TITLE}" --body-file "${BODY_FILE}" \
+              || echo "could not file the stall alert issue either; check the token" >&2
+            exit 1
+          fi
+
+      - name: Notify on failure
+        if: failure()
+        run: |
+          echo "RustFS upgrade compatibility test failed"
+          echo "From: ${{ inputs.from_url || inputs.from_version || 'release (default)' }}"
+          echo "To: ${{ inputs.to_url || inputs.to_version || 'nightly (R2 latest)' }}"
+          echo "See the uploaded report and logs for details."
@@ -17,7 +17,7 @@ jobs:
    runs-on: ubuntu-latest
    timeout-minutes: 10
    steps:
-      - uses: overtrue/repo-visuals-action@72f34d24769ff5d341956da2f23952594ef2f1e2 # v1.3.0
+      - uses: overtrue/repo-visuals-action@fd79cba437ecfac933d00a69add17eb95d3939c3 # v1.3.1
        with:
          github-token: ${{ github.token }}
          output-branch: star-history
@@ -0,0 +1,192 @@
+# Copyright 2024 RustFS Team
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#     http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+
+name: Targets Integration
+
+on:
+  pull_request:
+    branches: [main]
+    paths:
+      - ".github/actions/setup/**"
+      - ".github/workflows/targets-integration.yml"
+      - "crates/targets/**"
+      - "Cargo.lock"
+  schedule:
+    - cron: "17 2 * * *"
+      timezone: "Asia/Shanghai"
+  workflow_dispatch:
+
+permissions:
+  contents: read
+
+concurrency:
+  group: targets-integration-${{ github.ref }}-${{ github.event_name }}
+  cancel-in-progress: ${{ github.event_name != 'schedule' }}
+
+env:
+  CARGO_TERM_COLOR: always
+  RUST_BACKTRACE: 1
+
+jobs:
+  targets-live:
+    name: PostgreSQL, MySQL, AMQP, and NATS
+    runs-on: ubuntu-latest
+    timeout-minutes: 90
+    env:
+      FORCE_JAVASCRIPT_ACTIONS_TO_NODE24: "true"
+      NO_PROXY: 127.0.0.1,localhost
+      RUSTFS_TEST_PG_DSN: postgres://postgres:[email protected]:5432/rustfs_events
+      RUSTFS_TEST_MYSQL_DSN: root:testpass@tcp(127.0.0.1:3306)/testdb
+      RUSTFS_TEST_AMQP_URL: amqp://rustfs:[email protected]:5672/%2f
+      RUSTFS_TEST_NATS_URL: nats://127.0.0.1:4222
+    steps:
+      - name: Checkout repository
+        uses: actions/checkout@9c091bb21b7c1c1d1991bb908d89e4e9dddfe3e0 # v7
+        with:
+          persist-credentials: false
+
+      - name: Setup Rust environment
+        uses: ./.github/actions/setup
+        with:
+          cache-shared-key: targets-live-lane
+          cache-save-if: ${{ github.ref == 'refs/heads/main' || github.event_name == 'schedule' }}
+          install-build-packaging-tools: 'false'
+          install-test-tools: 'false'
+
+      - name: Start target services
+        run: |
+          set -euo pipefail
+          mkdir -p artifacts/targets-live/services
+          docker run -d --name rustfs-targets-postgres \
+            -e POSTGRES_PASSWORD=rustfs \
+            -e POSTGRES_DB=rustfs_events \
+            -p 5432:5432 postgres:16
+          docker run -d --name rustfs-targets-mysql \
+            -e MYSQL_ROOT_PASSWORD=testpass \
+            -e MYSQL_DATABASE=testdb \
+            -p 3306:3306 mysql:8.0.36
+          docker run -d --name rustfs-targets-rabbitmq \
+            -e RABBITMQ_DEFAULT_USER=rustfs \
+            -e RABBITMQ_DEFAULT_PASS=rustfs \
+            -p 5672:5672 rabbitmq:3
+          docker run -d --name rustfs-targets-nats \
+            -p 4222:4222 -p 8222:8222 nats:2 -js -m 8222
+
+          for _ in $(seq 1 120); do
+            docker exec rustfs-targets-postgres pg_isready -U postgres -d rustfs_events >/dev/null 2>&1 && break
+            sleep 1
+          done
+          docker exec rustfs-targets-postgres pg_isready -U postgres -d rustfs_events
+
+          for _ in $(seq 1 120); do
+            docker exec rustfs-targets-mysql mysqladmin ping -h 127.0.0.1 -uroot -ptestpass --silent >/dev/null 2>&1 && break
+            sleep 1
+          done
+          docker exec rustfs-targets-mysql mysqladmin ping -h 127.0.0.1 -uroot -ptestpass --silent
+
+          for _ in $(seq 1 120); do
+            docker exec rustfs-targets-rabbitmq rabbitmq-diagnostics -q ping >/dev/null 2>&1 && break
+            sleep 1
+          done
+          docker exec rustfs-targets-rabbitmq rabbitmq-diagnostics -q ping
+
+          for _ in $(seq 1 120); do
+            curl -fsS http://127.0.0.1:8222/healthz >/dev/null 2>&1 && break
+            sleep 1
+          done
+          curl -fsS http://127.0.0.1:8222/healthz
+
+      - name: Run live target tests
+        env:
+          CARGO_BUILD_JOBS: "2"
+        run: |
+          set +e
+          timeout --verbose --signal=TERM --kill-after=30s 75m bash <<'TESTS' \
+            2>&1 | tee artifacts/targets-live/tests.log
+          result=0
+
+          echo "::group::PostgreSQL"
+          cargo test --locked -p rustfs-targets --test postgres_integration -- --ignored --test-threads=1 || result=1
+          echo "::endgroup::"
+
+          echo "::group::MySQL"
+          cargo test --locked -p rustfs-targets --test mysql_integration -- --ignored --test-threads=1 || result=1
+          echo "::endgroup::"
+
+          echo "::group::AMQP"
+          cargo test --locked -p rustfs-targets --test amqp_integration -- --ignored --test-threads=1 || result=1
+          echo "::endgroup::"
+
+          echo "::group::NATS integration"
+          cargo test --locked -p rustfs-targets --test nats_jetstream_validation_integration -- --ignored --test-threads=1 || result=1
+          cargo test --locked -p rustfs-targets --test nats_jetstream_regression_guards -- --ignored --test-threads=1 || result=1
+          cargo test --locked -p rustfs-targets --lib target::nats::jetstream -- --ignored --test-threads=1 || result=1
+          echo "::endgroup::"
+
+          exit "${result}"
+          TESTS
+          status=${PIPESTATUS[0]}
+          {
+            echo "exit_status=${status}"
+            echo "finished_at=$(date --utc --iso-8601=seconds)"
+            echo
+            echo "Remaining test-related processes:"
+            pgrep -af 'cargo|target/.*/deps/' || true
+          } > artifacts/targets-live/diagnostics.txt
+          exit "${status}"
+
+      - name: Collect service logs
+        if: always()
+        run: |
+          mkdir -p artifacts/targets-live/services
+          for container in postgres mysql rabbitmq nats; do
+            docker logs --tail 500 "rustfs-targets-${container}" \
+              > "artifacts/targets-live/services/${container}.log" 2>&1 || true
+          done
+
+      - name: Stop target services
+        if: always()
+        run: |
+          docker rm -f \
+            rustfs-targets-postgres \
+            rustfs-targets-mysql \
+            rustfs-targets-rabbitmq \
+            rustfs-targets-nats >/dev/null 2>&1 || true
+
+      - name: Upload target integration diagnostics
+        if: always()
+        uses: actions/upload-artifact@b7c566a772e6b6bfb58ed0dc250532a479d7789f # v6
+        with:
+          name: targets-integration-${{ github.run_number }}-${{ github.run_attempt }}
+          path: artifacts/targets-live
+
+  alert-on-failure:
+    name: Alert on scheduled failure
+    needs: [targets-live]
+    if: >-
+      always() && github.event_name == 'schedule' &&
+      (contains(needs.*.result, 'failure') || contains(needs.*.result, 'cancelled'))
+    runs-on: ubuntu-latest
+    timeout-minutes: 10
+    permissions:
+      contents: read
+      issues: write
+    steps:
+      - uses: actions/checkout@9c091bb21b7c1c1d1991bb908d89e4e9dddfe3e0 # v7
+        with:
+          persist-credentials: false
+      - name: Open or update failure-tracking issue
+        uses: ./.github/actions/schedule-failure-issue
+        with:
+          github-token: ${{ secrets.GITHUB_TOKEN }}
@@ -57,9 +57,6 @@ docs/*
 !docs/operations/**
 !docs/testing/
 !docs/testing/**
-docs/heal-scanner-logging-governance.md
-docs/benchmark/rustfs-target-bench/
-docs/benchmark/*.md
 .codegraph/*
 .docker/test/compat/data/*
 .docker/test/compat/kms/*
@@ -83,6 +80,8 @@ worktrees/*

 # Local AI-agent review artifacts (omo evidence dumps)
 .omo/
+# Legacy per-tool skill dir; skills live in .agents/skills (shared by all agents)
+.mimocode/

 # insta scratch files; the accepted .snap files ARE the assertions and are committed
 *.snap.new
@@ -31,8 +31,13 @@ This file contains repository-wide rules. Use the nearest subdirectory
 - An existing clean, isolated task worktree is sufficient. Create another
  worktree only when the current checkout is shared, dirty with unrelated work,
  or belongs to another task.
- Never commit from a shared checkout. Use an `overtrue/` feature branch unless
-  the user requests another name.
+- Never commit from a shared checkout.
+- Use a task-specific branch named `<type>/<topic>`, such as `fix/...`,
+  `feat/...`, `test/...`, or `docs/...`, unless the user specifies a name.
+- Do not include agent, tool, contributor, account, or organization names in
+  branch names.
+- Push to the user-requested remote or the repository's configured push remote.
+  Do not hard-code or infer a remote from an account name.
 - Check free space before artifact-heavy builds, tests, coverage, or downloads.
  Re-check before a broad gate when space is tight.
 - Remove only task-owned temporary/build artifacts. Never delete another task's
@@ -81,6 +86,7 @@ This file contains repository-wide rules. Use the nearest subdirectory
 - CI gates: `.github/workflows/ci.yml`.
 - PR format: `.github/pull_request_template.md`.
 - Architecture routing: `ARCHITECTURE.md` and `docs/architecture/README.md`.
+- Knowledge-base index and documentation rules: `docs/architecture/README.md`.
 - Agent skills: `.agents/skills/*/SKILL.md`.

 Do not commit one-shot plans, trackers, migration ledgers, benchmark snapshots,
@@ -118,12 +124,13 @@ runtime/build output:
 - Use `make pre-commit` only when its repository-wide fast checks add confidence
  beyond the focused checks.

-### Broad or High-Risk Changes
+### Broad Cross-Module Changes

-After the required adversarial review, run `make pre-pr` when targeted coverage
-cannot bound the impact, including dependency/toolchain/build-matrix changes,
-unbounded cross-crate APIs, or locking, durability, erasure coding, replication,
-RPC, IAM/KMS/auth, cryptography, on-disk/on-wire, and S3-visible behavior.
+Do not run `make pre-pr` by default before opening a PR. Consider it only when
+the final diff is broad, spans multiple modules, and targeted checks cannot
+bound the impact. Decide dynamically from the affected boundaries and risks;
+otherwise use the scoped formatting, linting, compilation, and test checks
+above.

 `make pre-pr` includes `make pre-commit`; never run both for the same unchanged
 diff. Do not repeat a check already covered by a successful umbrella gate.
@@ -62,7 +62,7 @@ rustfs/                      # Workspace root (virtual manifest)
 │   ├── utils/               # Pure utility functions
 │   ├── ...                  # (see "Crate Reference" below)
 │   └── e2e_test/            # End-to-end integration tests
-└── docs/                    # Design documents and analysis
+└── docs/                    # Agent knowledge base: contracts, runbooks, testing rules (index: docs/architecture/README.md)
 ```

 ### Main Crate Layers (`rustfs/src/`)
@@ -73,7 +73,7 @@ The main crate is organized in layers, top to bottom:
 |-------|-----------|----------------|
 | **Server** | `server/` | HTTP listener, TLS, CORS, compression, middleware, graceful shutdown |
 | **Admin** | `admin/` | Admin API routing, 30+ handler modules, web console |
-| **App** | `app/` | Use-case orchestration: object_usecase, bucket_usecase, multipart_usecase |
+| **App** | `app/` | Use-case orchestration: object (per-operation modules under `app/object/`, re-exported as `object_usecase`), bucket_usecase, multipart_usecase |
 | **Storage** | `storage/` | S3 API translation, erasure-coded FS, SSE encryption, RPC, concurrency |
 | **Auth** | `auth.rs` | S3 signature verification, credential validation |
 | **Config** | `config/` | CLI parsing, config struct, workload profiles |
@@ -92,8 +92,8 @@ refactors.

 | Domain | Current workspace crates | Responsibility |
 |--------|--------------------------|----------------|
-| Foundation | `checksums`, `common`, `config`, `data-usage`, `utils` | Shared configuration, data-usage models, utilities, and checksums. |
-| I/O and storage | `concurrency`, `ecstore`, `filemeta`, `heal`, `io-core`, `io-metrics`, `lifecycle`, `lock`, `object-capacity`, `object-data-cache`, `replication`, `rio`, `rio-v2`, `scanner`, `storage-api` | Erasure-coded object storage, metadata, recovery, lifecycle, replication, locking, cache, and I/O pipelines. |
+| Foundation | `checksums`, `common`, `config`, `data-usage`, `heal-contracts`, `scanner-contracts`, `utils` | Shared configuration, data-usage models, heal/scanner domain contracts, utilities, and checksums. |
+| I/O and storage | `concurrency`, `ecstore`, `filemeta`, `heal`, `io-core`, `io-metrics`, `lifecycle`, `lock`, `object-capacity`, `object-data-cache`, `replication`, `rio`, `rio-v2`, `s3-client`, `scanner`, `storage-api` | Erasure-coded object storage, metadata, recovery, lifecycle, replication, locking, cache, I/O pipelines, and the engine-side S3 client for remote tier/transition targets. |
 | Security and identity | `credentials`, `crypto`, `iam`, `keystone`, `kms`, `policy`, `security-governance`, `signer`, `tls-runtime`, `trusted-proxies` | Credentials, authentication, authorization, encryption, key management, TLS, and security contracts. |
 | Protocols and contracts | `extension-schema`, `madmin`, `protos`, `protocols`, `s3-ops`, `s3-types`, `s3select-api`, `s3select-query` | Admin, inter-node, S3, S3 Select, and optional protocol contracts. |
 | Operations and integration | `audit`, `notify`, `obs`, `targets`, `zip` | Auditing, observability, event delivery, notification targets, and archive support. |
@@ -115,8 +115,15 @@ default build (lifecycle:
 1. **Layers flow downward.** Server → Admin/App → Storage → ecstore → rio/io-core.
   No upward imports.

-2. **Leaf crates have zero internal dependencies.** `config`, `credentials`, `crypto`,
-   `io-metrics`, and `madmin` should depend only on external crates.
+2. **Leaf crates depend only on external crates, with adjudicated exceptions
+   pinned by a guard.** `config`, `credentials`, and `crypto` take no internal
+   dependency. `io-metrics` takes exactly `rustfs-s3-ops` (transitively
+   `rustfs-s3-types`), a pure contract crate with no I/O and no global state —
+   adjudicated in rustfs/backlog#1834. `madmin` left the leaf set when #6166 made
+   it the SigV4-signed admin SDK client; its internal dependency surface is pinned
+   to exactly `rustfs-signer`. Both pins live in the leaf allowlist in
+   `scripts/check_architecture_migration_rules.sh`; any other internal dependency
+   fails the guard ([crate boundaries](docs/architecture/crate-boundaries.md)).
   - ✅ RESOLVED: the historical `utils → config` and `common → filemeta`/`madmin`
     edges were removed; do not reintroduce them (see Known Structural Issues).

@@ -128,7 +135,7 @@ default build (lifecycle:
     `crates/ecstore/src/bucket/replication/replication_state.rs`) — a naming
     collision, not copies; renaming is tracked in rustfs/backlog#1847.
   - `LastMinuteLatency` has two deliberately different implementations: the
-     per-second bucketed accumulator in `crates/common/src/last_minute.rs` and
+     per-second bucketed accumulator in `crates/scanner-contracts/src/last_minute.rs` and
     the in-memory endpoint-health sample tracker in
     `crates/ecstore/src/bucket/bucket_target_sys.rs` (its doc comment explains
     why it stays local).
@@ -138,15 +145,19 @@ default build (lifecycle:
     `BackpressureSettings` copy that lingered in io-metrics was removed
     (rustfs/backlog#1833).

-4. **ecstore does not know about HTTP or S3 protocol details.** It operates on
-   storage-level abstractions (objects, buckets, disks, pools).
-   - ⚠️ VIOLATED: 58 files under `crates/ecstore/src` reference `s3s`
-     (`rg -l 's3s' crates/ecstore/src | wc -l`), `crates/ecstore/src/client/`
-     is a ~9.4K-line embedded S3 HTTP client, and `crates/ecstore/Cargo.toml`
-     depends on `s3s`, `http`, `hyper`/`hyper-util`/`hyper-rustls`, and
-     `reqwest`. Target state: the engine's need to act as an S3 client
-     (tiering, replication targets) is served by an extracted client crate,
-     and ecstore holds no wire or DTO types.
+4. **ecstore does not *serve* HTTP or the S3 wire protocol.** It operates on
+   storage-level abstractions (objects, buckets, disks, pools) and holds no
+   wire or DTO types of the serving surface. *Consuming* remote S3-compatible
+   endpoints (ILM tier warm backends, transition targets) is a legitimate
+   engine capability, but it lives in the dedicated `rustfs-s3-client` crate
+   (`crates/s3-client`, extracted from the formerly embedded
+   `crates/ecstore/src/client/` by rustfs/backlog#1842), not inside ecstore.
+   - ⚠️ PARTIALLY VIOLATED: serving-side `s3s` references remain in ecstore
+     (bucket metadata/replication/lifecycle DTOs and error mapping). The
+     count is ratcheted shrink-only by `scripts/check_s3s_footprint.sh`
+     (`S3S_ECSTORE_FILES_BASELINE`; the `object_lock` module was converted to
+     storage-level types as the first ratchet step). Target state: the
+     baseline reaches zero and ecstore's `Cargo.toml` drops `s3s`.

 5. **The `rustfs` binary crate is the only place that wires everything together.**
   Individual crates should be testable in isolation.
@@ -321,6 +332,8 @@ The binary (`main.rs`) boots in this order:

 - **"Where is replication configured?"**
  `admin/handlers/replication.rs` and `admin/handlers/site_replication.rs` for API,
+  `rustfs/src/site_replication/` for the site-replication service subsystem
+  (state, peer transport, retry queue, repair, hooks),
  `ecstore/src/bucket/replication/` for engine

 - **"Where do I add a new admin endpoint?"**
@@ -15,7 +15,7 @@ cargo check -p <crate>               # fast type-check one crate
 cargo test -p <crate>                # test one crate
 cargo fmt --all                      # format (required before PR)
 make pre-commit                      # fast gate: fmt + arch checks + quick-check (NO clippy/tests)
-make pre-pr                          # full pre-PR gate: fmt + arch checks + clippy + tests
+make pre-pr                          # optional full gate for broad cross-module changes
 make build-docker BUILD_OS=ubuntu22.04
 ```

@@ -27,6 +27,7 @@ make build-docker BUILD_OS=ubuntu22.04

 ## Where to look (do not duplicate here)

+- Agent knowledge base index and doc-writing rules: [docs/architecture/README.md](docs/architecture/README.md)
 - Crate membership: `Cargo.toml` `[workspace].members`
 - Architecture, layering, crate map: [ARCHITECTURE.md](ARCHITECTURE.md)
 - Migration guardrails & readiness contracts: [docs/architecture/](docs/architecture/README.md)
@@ -62,12 +62,20 @@ make test
 # Fast pre-commit gate — see below for exactly what it runs
 make pre-commit

-# Full pre-PR gate (pre-commit gates + clippy + tests)
+# Optional full gate for broad cross-module changes (pre-commit + clippy + tests)
 make pre-pr
 ```

 > `make test` requires [cargo-nextest](https://nexte.st) (CI runs it and only nextest honours `.config/nextest.toml` test-groups). Install it with `cargo install cargo-nextest --locked` or a prebuilt binary (see https://nexte.st/docs/installation/). To run the plain `cargo test` fallback anyway (results not authoritative — serialization semantics differ from CI), set `RUSTFS_ALLOW_CARGO_TEST_FALLBACK=1`.

+> Some guard checks are Python (`test-wiring-check` in `make pre-commit`, plus the
+> security-coverage and scheduled-validation self-tests in `make test`) and import
+> `tomllib`, so they need **Python 3.11+**. Make resolves the interpreter through
+> `scripts/python_bin.sh`, which prefers a `python3.11`+ on `PATH` and otherwise falls
+> back to `uv run --python 3.12`. macOS ships `/usr/bin/python3` at 3.9, so install a
+> newer one (`brew install [email protected]`) or [uv](https://docs.astral.sh/uv/); pin a
+> specific interpreter with `RUSTFS_PYTHON=/path/to/python3.12`.
+
 > For the full test-layer taxonomy (unit / ecstore black-box / e2e / s3s-e2e / S3 compatibility / chaos / fuzz / bench), each layer's entry command, the naming conventions the migration gate depends on, and the serial/nextest rules, see [docs/testing/README.md](docs/testing/README.md).

 > For the event, timeout, required-status, and local reproduction matrix, see [docs/testing/ci-gates.md](docs/testing/ci-gates.md).
@@ -88,14 +96,16 @@ make pre-pr
 8. `quick-check` — `cargo check --workspace --exclude e2e_test`

 **`make pre-commit` does NOT run clippy and does NOT run any tests.**
-A green `make pre-commit` is not enough to open a pull request.
+It does not replace the scoped Clippy and test checks applicable to a change.

 `make pre-pr` is the **full** gate: it runs all of the guard checks above,
 then `clippy-check` (`cargo clippy --all-targets --all-features -- -D warnings`)
 and `test` (shell script tests, workspace tests excluding `e2e_test`, and doc
 tests). Complete the applicable multi-role adversarial review described in
-`AGENTS.md` before running `make pre-pr`; then run the gate before opening or
-updating a pull request. This is what CI enforces.
+`AGENTS.md` first. Do not run `make pre-pr` locally by default before opening or
+updating a pull request. Consider it only for a broad change that spans multiple
+modules and whose impact cannot be bounded by targeted checks; decide from the
+affected boundaries and risks. CI still runs its configured repository gates.

 ### 🔒 Git Pre-commit Hooks (optional)

@@ -114,8 +124,9 @@ Or manually:
 chmod +x .git/hooks/pre-commit
 ```

-With or without a hook, the expectation is the same: run `make pre-commit`
-before committing and `make pre-pr` before opening a pull request.
+With or without a hook, follow the verification tiers in `AGENTS.md`. Run the
+applicable scoped checks, and reserve `make pre-pr` for broad cross-module
+changes whose impact cannot be bounded by those checks.

 ### 📝 Formatting Configuration

@@ -154,7 +165,9 @@ Example output when formatting fails:
 3. **Run the fast gate**: `make pre-commit` (no clippy, no tests)
 4. **Commit your changes**: `git commit -m "your message"`
 5. **Complete the applicable multi-role adversarial review** for non-exempt changes (see `AGENTS.md`)
-6. **Run the full gate before opening/updating a PR**: `make pre-pr` (clippy + tests)
+6. **Run applicable scoped checks before opening/updating a PR**; consider
+   `make pre-pr` only for broad cross-module changes whose impact cannot be
+   bounded by targeted checks
 7. **Push to your branch**: `git push`

 ### 🛠️ IDE Integration
@@ -26,8 +26,10 @@ members = [
    "crates/e2e_test", # End-to-end test suite
    "crates/filemeta", # File metadata management
    "crates/heal", # Erasure set and object healing
+    "crates/heal-contracts", # Heal request/response channel contracts
    "crates/iam", # Identity and Access Management
    "crates/keystone", # OpenStack Keystone integration
+    "crates/license", # License and entitlement provider contracts
    "crates/lifecycle", # Lifecycle rule evaluation contracts
    "crates/kms", # Key Management Service
    "crates/lock", # Distributed locking implementation
@@ -44,11 +46,13 @@ members = [
    "crates/rio-v2", # MinIO on-disk format compatibility I/O layer (feature-gated, ships in no default build)
    "crates/replication", # Replication contracts and wire formats
    "crates/concurrency", # Concurrency management for RustFS - timeout, locking, backpressure, and I/O scheduling
+    "crates/s3-client", # S3 client for engine-side consumption of remote S3 endpoints (tiering, transition targets)
    "crates/s3-types", # S3 event type definitions
    "crates/s3-ops", # S3 operation definitions and mapping
    "crates/s3select-api", # S3 Select API interface
    "crates/s3select-query", # S3 Select query engine
    "crates/scanner", # Scanner for data integrity checks and health monitoring
+    "crates/scanner-contracts", # Scanner metrics and cycle contracts
    "crates/security-governance", # Security governance contracts
    "crates/extension-schema", # Extension schema contracts
    "crates/signer", # client signer
@@ -68,8 +72,8 @@ resolver = "3"
 edition = "2024"
 license = "Apache-2.0"
 repository = "https://github.com/rustfs/rustfs"
-rust-version = "1.97.1"
-version = "1.0.0-rc.3"
+rust-version = "1.98.0"
+version = "1.0.0-rc.5"
 homepage = "https://rustfs.com"
 description = "RustFS is a high-performance distributed object storage software built using Rust, one of the most popular languages worldwide. "
 keywords = ["RustFS", "Minio", "object-storage", "filesystem", "s3"]
@@ -86,57 +90,61 @@ redundant_clone = "warn"

 [workspace.dependencies]
 # RustFS Internal Crates
-rustfs = { path = "./rustfs", version = "1.0.0-rc.3" }
-rustfs-heal = { path = "crates/heal", version = "1.0.0-rc.3" }
-rustfs-audit = { path = "crates/audit", version = "1.0.0-rc.3" }
-rustfs-checksums = { path = "crates/checksums", version = "1.0.0-rc.3" }
-rustfs-common = { path = "crates/common", version = "1.0.0-rc.3" }
-rustfs-data-usage = { path = "crates/data-usage", version = "1.0.0-rc.3" }
-rustfs-config = { path = "./crates/config", version = "1.0.0-rc.3" }
-rustfs-concurrency = { path = "./crates/concurrency", version = "1.0.0-rc.3" }
-rustfs-credentials = { path = "crates/credentials", version = "1.0.0-rc.3" }
-rustfs-crypto = { path = "crates/crypto", version = "1.0.0-rc.3" }
-rustfs-ecstore = { path = "crates/ecstore", version = "1.0.0-rc.3" }
-rustfs-filemeta = { path = "crates/filemeta", version = "1.0.0-rc.3" }
-rustfs-iam = { path = "crates/iam", version = "1.0.0-rc.3" }
-rustfs-keystone = { path = "crates/keystone", version = "1.0.0-rc.3" }
-rustfs-lifecycle = { path = "crates/lifecycle", version = "1.0.0-rc.3" }
-rustfs-kms = { path = "crates/kms", version = "1.0.0-rc.3" }
-rustfs-lock = { path = "crates/lock", version = "1.0.0-rc.3" }
-rustfs-madmin = { path = "crates/madmin", version = "1.0.0-rc.3" }
-rustfs-notify = { path = "crates/notify", version = "1.0.0-rc.3" }
-rustfs-io-metrics = { path = "crates/io-metrics", version = "1.0.0-rc.3" }
-rustfs-io-core = { path = "crates/io-core", version = "1.0.0-rc.3" }
-rustfs-object-capacity = { path = "crates/object-capacity", version = "1.0.0-rc.3" }
-rustfs-object-data-cache = { path = "crates/object-data-cache", version = "1.0.0-rc.3", default-features = false }
-rustfs-log-analyzer = { path = "crates/log-analyzer", version = "1.0.0-rc.3" }
-rustfs-obs = { path = "crates/obs", version = "1.0.0-rc.3" }
-rustfs-policy = { path = "crates/policy", version = "1.0.0-rc.3" }
-rustfs-protos = { path = "crates/protos", version = "1.0.0-rc.3" }
-rustfs-protocols = { path = "crates/protocols", version = "1.0.0-rc.3" }
-rustfs-replication = { path = "crates/replication", version = "1.0.0-rc.3" }
-rustfs-rio = { path = "crates/rio", version = "1.0.0-rc.3" }
-rustfs-rio-v2 = { path = "crates/rio-v2", version = "1.0.0-rc.3" }
-rustfs-s3-types = { path = "crates/s3-types", version = "1.0.0-rc.3" }
-rustfs-s3-ops = { path = "crates/s3-ops", version = "1.0.0-rc.3" }
-rustfs-s3select-api = { path = "crates/s3select-api", version = "1.0.0-rc.3" }
-rustfs-s3select-query = { path = "crates/s3select-query", version = "1.0.0-rc.3" }
-rustfs-scanner = { path = "crates/scanner", version = "1.0.0-rc.3" }
-rustfs-security-governance = { path = "crates/security-governance", version = "1.0.0-rc.3" }
-rustfs-extension-schema = { path = "crates/extension-schema", version = "1.0.0-rc.3" }
-rustfs-signer = { path = "crates/signer", version = "1.0.0-rc.3" }
-rustfs-storage-api = { path = "crates/storage-api", version = "1.0.0-rc.3" }
-rustfs-trusted-proxies = { path = "crates/trusted-proxies", version = "1.0.0-rc.3" }
-rustfs-targets = { path = "crates/targets", version = "1.0.0-rc.3" }
-rustfs-test-utils = { path = "crates/test-utils", version = "1.0.0-rc.3" }
-rustfs-tls-runtime = { path = "crates/tls-runtime", version = "1.0.0-rc.3" }
-rustfs-utils = { path = "crates/utils", version = "1.0.0-rc.3" }
-rustfs-zip = { path = "./crates/zip", version = "1.0.0-rc.3" }
+rustfs = { path = "./rustfs", version = "1.0.0-rc.5" }
+rustfs-heal = { path = "crates/heal", version = "1.0.0-rc.5" }
+rustfs-heal-contracts = { path = "crates/heal-contracts", version = "1.0.0-rc.5" }
+rustfs-scanner-contracts = { path = "crates/scanner-contracts", version = "1.0.0-rc.5" }
+rustfs-audit = { path = "crates/audit", version = "1.0.0-rc.5" }
+rustfs-checksums = { path = "crates/checksums", version = "1.0.0-rc.5" }
+rustfs-common = { path = "crates/common", version = "1.0.0-rc.5" }
+rustfs-data-usage = { path = "crates/data-usage", version = "1.0.0-rc.5" }
+rustfs-config = { path = "./crates/config", version = "1.0.0-rc.5" }
+rustfs-concurrency = { path = "./crates/concurrency", version = "1.0.0-rc.5" }
+rustfs-credentials = { path = "crates/credentials", version = "1.0.0-rc.5" }
+rustfs-crypto = { path = "crates/crypto", version = "1.0.0-rc.5" }
+rustfs-ecstore = { path = "crates/ecstore", version = "1.0.0-rc.5" }
+rustfs-filemeta = { path = "crates/filemeta", version = "1.0.0-rc.5" }
+rustfs-iam = { path = "crates/iam", version = "1.0.0-rc.5" }
+rustfs-keystone = { path = "crates/keystone", version = "1.0.0-rc.5" }
+rustfs-license = { path = "crates/license", version = "1.0.0-rc.5" }
+rustfs-lifecycle = { path = "crates/lifecycle", version = "1.0.0-rc.5" }
+rustfs-kms = { path = "crates/kms", version = "1.0.0-rc.5" }
+rustfs-lock = { path = "crates/lock", version = "1.0.0-rc.5" }
+rustfs-madmin = { path = "crates/madmin", version = "1.0.0-rc.5" }
+rustfs-notify = { path = "crates/notify", version = "1.0.0-rc.5" }
+rustfs-io-metrics = { path = "crates/io-metrics", version = "1.0.0-rc.5" }
+rustfs-io-core = { path = "crates/io-core", version = "1.0.0-rc.5" }
+rustfs-object-capacity = { path = "crates/object-capacity", version = "1.0.0-rc.5" }
+rustfs-object-data-cache = { path = "crates/object-data-cache", version = "1.0.0-rc.5", default-features = false }
+rustfs-log-analyzer = { path = "crates/log-analyzer", version = "1.0.0-rc.5" }
+rustfs-obs = { path = "crates/obs", version = "1.0.0-rc.5" }
+rustfs-policy = { path = "crates/policy", version = "1.0.0-rc.5" }
+rustfs-protos = { path = "crates/protos", version = "1.0.0-rc.5" }
+rustfs-protocols = { path = "crates/protocols", version = "1.0.0-rc.5" }
+rustfs-replication = { path = "crates/replication", version = "1.0.0-rc.5" }
+rustfs-rio = { path = "crates/rio", version = "1.0.0-rc.5" }
+rustfs-rio-v2 = { path = "crates/rio-v2", version = "1.0.0-rc.5" }
+rustfs-s3-client = { path = "crates/s3-client", version = "1.0.0-rc.5" }
+rustfs-s3-types = { path = "crates/s3-types", version = "1.0.0-rc.5" }
+rustfs-s3-ops = { path = "crates/s3-ops", version = "1.0.0-rc.5" }
+rustfs-s3select-api = { path = "crates/s3select-api", version = "1.0.0-rc.5" }
+rustfs-s3select-query = { path = "crates/s3select-query", version = "1.0.0-rc.5" }
+rustfs-scanner = { path = "crates/scanner", version = "1.0.0-rc.5" }
+rustfs-security-governance = { path = "crates/security-governance", version = "1.0.0-rc.5" }
+rustfs-extension-schema = { path = "crates/extension-schema", version = "1.0.0-rc.5" }
+rustfs-signer = { path = "crates/signer", version = "1.0.0-rc.5" }
+rustfs-storage-api = { path = "crates/storage-api", version = "1.0.0-rc.5" }
+rustfs-trusted-proxies = { path = "crates/trusted-proxies", version = "1.0.0-rc.5" }
+rustfs-targets = { path = "crates/targets", version = "1.0.0-rc.5" }
+rustfs-test-utils = { path = "crates/test-utils", version = "1.0.0-rc.5" }
+rustfs-tls-runtime = { path = "crates/tls-runtime", version = "1.0.0-rc.5" }
+rustfs-utils = { path = "crates/utils", version = "1.0.0-rc.5" }
+rustfs-zip = { path = "./crates/zip", version = "1.0.0-rc.5" }

 # Async Runtime and Networking
 async-channel = "2.5.0"
 async_zip = { default-features = false, version = "0.0.19" }
-mysql_async = { default-features = false, version = "0.37" }
+mysql_async = { default-features = false, version = "0.37.1" }
 async-compression = { version = "0.4.43" }
 async-recursion = "1.1.1"
 async-trait = "0.1.92"
@@ -147,9 +155,9 @@ futures-core = "0.3.34"
 futures-lite = "2.6.1"
 futures-util = "0.3.34"
 pollster = "1.0.1"
-pulsar = { default-features = false, version = "6.8.0" }
+pulsar = { default-features = false, version = "6.9.0" }
 lapin = { default-features = false, version = "4.10.0" }
-hyper = { version = "1.11.0" }
+hyper = { version = "1.11.1" }
 hyper-rustls = { default-features = false, version = "0.27.9" }
 hyper-util = { version = "0.1.20" }
 http = "1.5.0"
@@ -168,7 +176,7 @@ tonic = { version = "0.14.6" }
 tonic-prost = { version = "0.14.6" }
 tonic-prost-build = { version = "0.14.6" }
 tower = { version = "0.5.3" }
-tower-http = { version = "0.7.0" }
+tower-http = { version = "0.7.1" }

 # Serialization and Data Formats
 apache-avro = { version = "0.22.0", features = ["snappy", "zstandard"] }
@@ -192,14 +200,15 @@ serde_urlencoded = "0.7.1"
 # have incompatible APIs. Keep them exact-pinned and monitor upstream for stable
 # releases.
 aes-gcm = { version = "=0.11.1" }
-argon2 = { version = "=0.6.0-rc.8" }
-blake2 = "=0.11.0-rc.6"
+argon2 = { version = "=0.6.0" }
+blake2 = "=0.11.0"
 chacha20poly1305 = { version = "=0.11.0" }
 crc-fast = "1.10.0"
 hmac = { version = "0.13.0" }
 jsonwebtoken = { version = "11.0.0" }
 openidconnect = { default-features = false, version = "4.0" }
 pbkdf2 = "0.13.0"
+p256 = { version = "0.14.0", features = ["ecdsa", "pkcs8"] }
 rsa = { version = "=0.10.0-rc.18" }
 rustls = { default-features = false, version = "0.23.43" }
 rustls-native-certs = "0.8"
@@ -209,6 +218,7 @@ sha1 = "0.11.0"
 sha2 = "0.11.0"
 subtle = "2.6"
 zeroize = { version = "1.9.0" }
+proptest = "1"

 # Time and Date
 chrono = { version = "0.4.45" }
@@ -224,23 +234,23 @@ tokio-postgres-rustls = "0.14.0"
 # Utilities and Tools
 anyhow = "1.0.104"
 arc-swap = "1.9.2"
-astral-tokio-tar = "0.6.4"
+# RUSTFS_COMPAT_TODO(tokio-tar-extension-limits): keep the fork pin until every parser hardening used by Snowball is released upstream. Remove after astral-sh/tokio-tar#118 is merged and a published release includes extension, physical-entry, and sparse limits, cancellation-safe sparse parsing, and error-fused entry streams.
+astral-tokio-tar = { git = "https://github.com/cxymds/tokio-tar.git", rev = "603756478b7668436e464519c77ccac22a99ba96" }
 atoi = "3.1.0"
 atomic_enum = "0.3.0"
 aws-config = { version = "1.11.0" }
 aws-credential-types = { version = "1.3.0" }
-aws-sdk-kms = { default-features = false, version = "1.116.0" }
-aws-sdk-s3 = { default-features = false, version = "1.143.0" }
-aws-sdk-sts = { default-features = false, version = "1.112.0" }
+aws-sdk-kms = { default-features = false, version = "1.117.0" }
+aws-sdk-s3 = { default-features = false, version = "1.144.0" }
+aws-sdk-sts = { default-features = false, version = "1.113.0" }
 aws-smithy-http-client = { default-features = false, version = "1.4.0" }
 aws-smithy-runtime-api = { version = "1.15.0" }
 aws-smithy-types = { version = "1.6.2" }
-base64 = "0.23.1"
 base64-simd = "0.8.0"
-brotli = "8.0.4"
+brotli = "9.0.0"
 clap = { version = "4.6.6" }
 const-str = { version = "1.1.0" }
-convert_case = "0.11.0"
+convert_case = "0.12.0"
 criterion = { version = "0.8" }
 crossbeam-queue = "0.3.13"
 crossbeam-channel = "0.5.16"
@@ -250,12 +260,14 @@ datafusion = { default-features = false, version = "55.0.0" }
 derive_builder = "0.20.2"
 enumset = "1.1.14"
 faster-hex = "0.10.0"
-flate2 = "1.1.9"
+flate2 = "1.1.10"
 glob = "0.3.4"
-google-cloud-storage = "1.17.0"
-google-cloud-auth = "1.15.0"
+google-cloud-storage = "1.18.0"
+google-cloud-auth = "1.16.0"
 hashbrown = { version = "0.17.1" }
-hex = "0.4.3"
+# Base32 for RFC 6238 TOTP shared secrets (RFC 4648 unpadded, the alphabet
+# every authenticator app expects). Already in the graph transitively.
+data-encoding = "2.11.1"
 hex-simd = "0.8.0"
 highway = { version = "1.3.0" }
 hostname = "0.4.2"
@@ -273,10 +285,14 @@ mime_guess = "2.0.5"
 moka = { version = "0.12.16" }
 netif = "0.1.6"
 num_cpus = { version = "1.17.0" }
-nvml-wrapper = "0.12.1"
+nvml-wrapper = "0.13.0"
 parking_lot = "0.12.5"
 path-absolutize = "4.0.1"
 percent-encoding = "2.3.2"
+# Server-side QR rendering for TOTP enrollment, so neither the console nor the
+# CLI needs its own QR encoder. No default features: the image/render backends
+# pull in an image stack this only needs SVG and text output from.
+qrcode-rs = { version = "2.0.0", default-features = false, features = ["std", "svg"] }
 pin-project-lite = "0.2.17"
 pretty_assertions = "1.4.1"
 rand = { version = "0.10.2" }
@@ -291,14 +307,14 @@ rustify = { version = "0.7", default-features = false }
 rustix = { version = "1.1.4" }
 rust-embed = { version = "8.12.0" }
 rustc-hash = { version = "2.1.3" }
-s3s = { git = "https://github.com/rustfs/s3s.git", rev = "e080e38c56a3b43acbacce55710d765a5ce9003d" }
+s3s = { git = "https://github.com/rustfs/s3s.git", rev = "bdcb6259339c41369f9f1c60e3a42b5ab8da607b", version = "0.15.0", features = ["minio"] }
 serial_test = "4.0.1"
 shadow-rs = { default-features = false, version = "2.0.0" }
 siphasher = "1.0.3"
-smallvec = { version = "1.15.2" }
+smallvec = { version = "1.16.0" }
 compact_str = "0.10.0"
 snap = "1.1.2"
-starshard = { version = "2.2.2" }
+starshard = { version = "2.3.0" }
 strum = { version = "0.28.0" }
 sysinfo = "0.39.6"
 temp-env = "0.3.6"
@@ -314,7 +330,7 @@ tracing-subscriber = { version = "0.3.23" }
 transform-stream = "0.3.1"
 url = "2.5.8"
 urlencoding = "2.1.3"
-uuid = { version = "1.25.0" }
+uuid = { version = "1.26.0" }
 vaultrs = { version = "0.8.0" }
 tar = "0.4.46"
 walkdir = "2.5.0"
@@ -328,7 +344,7 @@ zstd = "0.13.3"
 # Observability and Metrics
 metrics = "0.24.6"
 metrics-util = "0.20"
-dial9-tokio-telemetry = "0.3"
+dial9-tokio-telemetry = "0.5.0"
 opentelemetry = { version = "0.32.0" }
 opentelemetry-appender-tracing = { version = "0.32.0" }
 opentelemetry-otlp = { version = "0.32.0" }
@@ -341,17 +357,16 @@ pyroscope = { version = "2.1.1" }
 # FTP and SFTP
 libunftp = { version = "0.23.0" }
 unftp-core = "0.1.0"
-suppaftp = { version = "10.0.2" }
-rcgen = { version = "0.14.9", default-features = false, features = ["aws_lc_rs", "crypto", "pem"] }
-russh = { version = "0.63.0" }
+suppaftp = { version = "11.0.0" }
+rcgen = { version = "0.14.10", default-features = false, features = ["aws_lc_rs", "crypto", "pem"] }
+russh = { version = "0.63.1" }
 russh-sftp = "2.4.0"

 # WebDAV
 dav-server = "0.11.0"

 # Performance Analysis and Memory Profiling
-rustfs-mimalloc = { version = "0.5.0" }
-rustfs-mimalloc-sys = { version = "0.5.0" }
+rustfs-mimalloc = { version = "0.5.1" }
 hotpath = { version = "0.24.0", default-features = false }
 # Snapshot testing for output format regression detection
 insta = { version = "1.48" }
@@ -371,8 +386,8 @@ opt-level = 3
 lto = "thin"
 codegen-units = 1
 debug = 0
-split-debuginfo = "off"
 strip = "symbols"
+split-debuginfo = "off"

 [profile.production]
 inherits = "release"
@@ -23,6 +23,12 @@ SHELL := $(shell which bash)
 .SHELLFLAGS = -eu -o pipefail -c

 DOCKER_CLI ?= docker
+# Python interpreter for the repository's helper scripts. They import tomllib
+# (Python 3.11+), while macOS still ships /usr/bin/python3 at 3.9, so calls go
+# through a resolver that picks a new-enough interpreter (or falls back to uv).
+# Override with RUSTFS_PYTHON=/path/to/python3.12, or replace the resolver via
+# RUSTFS_PYTHON_BIN=<command>.
+RUSTFS_PYTHON_BIN ?= ./scripts/python_bin.sh
 IMAGE_NAME ?= rustfs:v1.0.0
 CONTAINER_NAME ?= rustfs-dev
 # Docker build configurations
@@ -12,12 +12,11 @@
 </p>

 <p align="center">
-<a href="https://trendshift.io/repositories/14181" target="_blank"><img src="https://trendshift.io/api/badge/repositories/14181" alt="rustfs%2Frustfs | Trendshift" style="width: 250px; height: 55px;" width="250" height="55"/></a> 
-<a href="https://runacap.com/ross-index/q4-2025/" target="_blank" rel="noopener"><img style="width: 260px; height: 55px" src="https://runacap.com/wp-content/uploads/2026/01/ROSS_badge_white_Q4_2025.svg" alt="ROSS Index - Fastest Growing Open-Source Startups in Q4 2025 | Runa Capital" height="55" /></a>
+<a href="https://trendshift.io/repositories/14181" target="_blank"><img src="https://trendshift.io/api/badge/repositories/14181" alt="rustfs%2Frustfs | Trendshift" style="width: 250px; height: 55px;" width="250" height="55"/></a>
 </p>

 <p align="center">
-  <a href="https://docs.rustfs.com/installation/">Getting Started</a>
+  <a href="https://docs.rustfs.com/en/installation">Getting Started</a>
  · <a href="https://docs.rustfs.com/">Docs</a>
  · <a href="https://github.com/rustfs/rustfs/issues">Bug reports</a>
  · <a href="https://github.com/rustfs/rustfs/discussions">Discussions</a>
@@ -49,16 +48,33 @@ Unlike other storage systems, RustFS is released under the permissible Apache 2.
 - **Open Source**: Licensed under Apache 2.0, encouraging unrestricted community contributions and commercial usage.
 - **User-Friendly**: Designed with simplicity in mind for easy deployment and management.

-| Feature                 | Status       | Feature                  | Status           |
-| :---------------------- | :----------- | :----------------------- | :--------------- |
-| **S3 Core Features**    | ✅ Available | **Bitrot Protection**    | ✅ Available     |
-| **Upload / Download**   | ✅ Available | **Single Node Mode**     | ✅ Available     |
-| **Versioning**          | ✅ Available | **Bucket Replication**   | ✅ Available     |
-| **Logging**             | ✅ Available | **Lifecycle Management** | 🚧 Under Testing |
-| **Event Notifications** | ✅ Available | **Distributed Mode**     | 🚧 Under Testing |
-| **K8s Helm Charts**     | ✅ Available | **RustFS KMS**           | 🚧 Under Testing |
-| **Keystone Auth**       | ✅ Available | **Multi-Tenancy**        | ✅ Available     |
-| **Swift API**           | ✅ Available | **Swift Metadata Ops**   | 🚧 Partial       |
+Status legend: ✅ Available — shipped and covered by CI gates; 🧪 Preview — shipped behind an opt-in flag or with a bounded compatibility claim.
+
+| Feature                          | Status       | Feature                            | Status       |
+| :------------------------------- | :----------- | :--------------------------------- | :----------- |
+| **S3 Core Features**             | ✅ Available | **Distributed Mode**               | ✅ Available |
+| **Upload / Download**            | ✅ Available | **Single Node Mode**               | ✅ Available |
+| **Versioning**                   | ✅ Available | **Bitrot Protection**              | ✅ Available |
+| **Object Lock (WORM)**           | ✅ Available | **Healing & Scanner**              | ✅ Available |
+| **Server-Side Encryption**       | ✅ Available | **Pool Expansion / Decommission**  | ✅ Available |
+| **RustFS KMS**                   | ✅ Available | **Bucket Replication**             | ✅ Available |
+| **Lifecycle Management (ILM)**   | ✅ Available | **Site Replication**               | ✅ Available |
+| **ILM Tiering (Remote S3)**      | ✅ Available | **Bucket Quota**                   | ✅ Available |
+| **S3 Select**                    | ✅ Available | **Event Notifications**            | ✅ Available |
+| **S3 Tables (Iceberg REST)**     | 🧪 Preview   | **Audit Logging**                  | ✅ Available |
+| **IAM / Policies**               | ✅ Available | **Logging & Observability**        | ✅ Available |
+| **OIDC / SSO**                   | ✅ Available | **Web Console**                    | ✅ Available |
+| **Keystone Auth**                | ✅ Available | **K8s Helm Charts**                | ✅ Available |
+| **Swift API**                    | ✅ Available | **FTPS / WebDAV**                  | ✅ Available |
+| **Multi-Tenancy**                | ✅ Available | **SFTP**                           | ✅ Available |
+| **MinIO On-Disk Compatibility**  | 🧪 Preview   |                                    |              |
+
+Notes:
+
+- **RustFS KMS**: Vault (KV2 / Transit) and AWS KMS backends are supported for production. The `Local` and `Static` backends are for development and testing only. See [KMS backend security properties](docs/operations/kms-backend-security.md).
+- **Swift API / SFTP**: opt-in cargo features (`--features swift`, `--features sftp`, or `full`). FTPS and WebDAV are enabled in the default build.
+- **S3 Tables**: ships as an Iceberg REST Catalog with automated PyIceberg and DuckDB coverage; other engines and vendor profiles carry bounded claims listed in the [S3 Tables support matrix](docs/architecture/s3-tables-support-matrix.md).
+- **MinIO On-Disk Compatibility**: gated behind the `rio-v2` feature and not part of the default build. Objects MinIO encrypted are not readable by RustFS. See [MinIO file-format interoperability](docs/architecture/minio-file-format-compat.md).

 ## RustFS vs MinIO Performance

@@ -116,7 +132,7 @@ chown -R 10001:10001 data logs
 docker run -d -p 9000:9000 -p 9001:9001 -v $(pwd)/data:/data -v $(pwd)/logs:/logs rustfs/rustfs:latest

 # Using specific version
-docker run -d -p 9000:9000 -p 9001:9001 -v $(pwd)/data:/data -v $(pwd)/logs:/logs rustfs/rustfs:1.0.0-rc.3
+docker run -d -p 9000:9000 -p 9001:9001 -v $(pwd)/data:/data -v $(pwd)/logs:/logs rustfs/rustfs:1.0.0-rc.5
 ```

 If you use [podman](https://github.com/containers/podman) instead of docker, you can install the RustFS with the below command
@@ -246,6 +262,26 @@ nix build
 nix run
 ```

+The flake also exports a NixOS module and the RustFS `rc` client. Add the
+module to your system and provide credentials through runtime files (for
+example, sops-nix or agenix) so secrets are never stored in the Nix store:
+
+```nix
+imports = [ inputs.rustfs.nixosModules.rustfs ];
+
+services.rustfs = {
+  enable = true;
+  accessKeyFile = "/run/secrets/rustfs-access-key";
+  secretKeyFile = "/run/secrets/rustfs-secret-key";
+  volumes = [ "/var/lib/rustfs" ];
+};
+```
+
+Install the S3-compatible client with
+`nix profile install github:rustfs/rustfs#rustfs-client` (the executable is named
+`rc`), or use `inputs.rustfs.packages.${pkgs.system}.rustfs-client` in a system
+configuration.
+
 ### 6\. X-CMD (Option 6)

 If you are an [x-cmd](https://www.x-cmd.com/install/rustfs) user:
@@ -12,12 +12,11 @@


 <p align="center">
-<a href="https://trendshift.io/repositories/14181" target="_blank"><img src="https://trendshift.io/api/badge/repositories/14181" alt="rustfs%2Frustfs | Trendshift" style="width: 250px; height: 55px;" width="250" height="55"/></a> 
-<a href="https://runacap.com/ross-index/q4-2025/" target="_blank" rel="noopener"><img style="width: 260px; height: 55px" src="https://runacap.com/wp-content/uploads/2026/01/ROSS_badge_white_Q4_2025.svg" alt="ROSS Index - Fastest Growing Open-Source Startups in Q4 2025 | Runa Capital" height="55" /></a>
+<a href="https://trendshift.io/repositories/14181" target="_blank"><img src="https://trendshift.io/api/badge/repositories/14181" alt="rustfs%2Frustfs | Trendshift" style="width: 250px; height: 55px;" width="250" height="55"/></a>
 </p>

 <p align="center">
-  <a href="https://docs.rustfs.com/installation/">快速开始</a>
+  <a href="https://docs.rustfs.com/zh/installation">快速开始</a>
  · <a href="https://docs.rustfs.com/">文档</a>
  · <a href="https://github.com/rustfs/rustfs/issues">报告 Bug</a>
  · <a href="https://github.com/rustfs/rustfs/discussions">社区讨论</a>
@@ -113,7 +112,7 @@ chown -R 10001:10001 data logs
 docker run -d -p 9000:9000 -p 9001:9001 -v $(pwd)/data:/data -v $(pwd)/logs:/logs rustfs/rustfs:latest

 # 使用指定版本运行
-docker run -d -p 9000:9000 -p 9001:9001 -v $(pwd)/data:/data -v $(pwd)/logs:/logs rustfs/rustfs:1.0.0-rc.3
+docker run -d -p 9000:9000 -p 9001:9001 -v $(pwd)/data:/data -v $(pwd)/logs:/logs rustfs/rustfs:1.0.0-rc.5
 ```

 如果您通过绑定挂载启用 TLS 证书目录，也请用同样方式准备该目录：
@@ -192,6 +191,12 @@ nix build
 nix run
 ```

+该 Flake 同时提供 NixOS 模块和 RustFS `rc` 客户端。将
+`inputs.rustfs.nixosModules.rustfs` 加入 `imports`，并通过运行时密钥文件
+（例如 sops-nix 或 agenix）配置 `accessKeyFile` 与 `secretKeyFile`，避免密钥
+进入 Nix store。客户端包为
+`inputs.rustfs.packages.${pkgs.system}.rustfs-client`，安装后的命令名为 `rc`。
+
 ### 6\. X-CMD (Option 6)

 如果你是 [x-cmd](https://www.x-cmd.com/install/rustfs) 用户：
@@ -67,7 +67,7 @@ tokio = { workspace = true, features = ["sync", "fs", "rt-multi-thread", "time",
 tracing = { workspace = true, features = ["std", "attributes"] }

 [dev-dependencies]
-async-trait = { workspace = true }
+rustfs-targets = { workspace = true, features = ["test-support"] }
 temp-env = { workspace = true }
 url = { workspace = true }

@@ -564,88 +564,21 @@ impl AuditRuntimeFacade {
 mod tests {
    use super::AuditPipeline;
    use crate::{AuditEntry, AuditError, AuditRegistry};
-    use async_trait::async_trait;
-    use rustfs_targets::arn::TargetID;
-    use rustfs_targets::store::{Key, Store};
-    use rustfs_targets::target::{EntityTarget, QueuedPayload, QueuedPayloadMeta};
-    use rustfs_targets::{StoreError, Target, TargetError};
+    use rustfs_targets::testkit::MockTarget;
    use std::sync::Arc;
    use tokio::sync::{Mutex, Notify};

-    /// Mock target whose `save()` outcome is fixed at construction so tests can
-    /// force full-success / full-failure / partial-failure fan-outs.
-    #[derive(Clone)]
-    struct MockTarget {
-        id: TargetID,
-        fail: bool,
-        health_gate: Option<(Arc<Notify>, Arc<Notify>)>,
-    }
-
-    impl MockTarget {
-        fn new(id: &str, fail: bool) -> Self {
-            Self {
-                id: TargetID::new(id.to_string(), "webhook".to_string()),
-                fail,
-                health_gate: None,
-            }
-        }
-
-        fn with_health_gate(mut self, started: Arc<Notify>, release: Arc<Notify>) -> Self {
-            self.health_gate = Some((started, release));
-            self
-        }
-    }
-
-    #[async_trait]
-    impl<E> Target<E> for MockTarget
-    where
-        E: rustfs_targets::PluginEvent,
-    {
-        fn id(&self) -> TargetID {
-            self.id.clone()
-        }
-
-        async fn is_active(&self) -> Result<bool, TargetError> {
-            if let Some((started, release)) = &self.health_gate {
-                started.notify_one();
-                release.notified().await;
-            }
-            Ok(true)
-        }
-
-        async fn save(&self, _event: Arc<EntityTarget<E>>) -> Result<(), TargetError> {
-            if self.fail {
-                Err(TargetError::Configuration("forced save failure".to_string()))
-            } else {
-                Ok(())
-            }
-        }
-
-        async fn send_raw_from_store(&self, _key: Key, _body: Vec<u8>, _meta: QueuedPayloadMeta) -> Result<(), TargetError> {
-            Ok(())
-        }
-
-        async fn close(&self) -> Result<(), TargetError> {
-            Ok(())
-        }
-
-        fn store(&self) -> Option<&(dyn Store<QueuedPayload, Error = StoreError, Key = Key> + Send + Sync)> {
-            None
-        }
-
-        fn clone_dyn(&self) -> Box<dyn Target<E> + Send + Sync> {
-            Box::new(self.clone())
-        }
-
-        fn is_enabled(&self) -> bool {
-            true
-        }
+    /// Builds a mock target whose `save()` outcome is fixed at construction so tests can force
+    /// full-success / full-failure / partial-failure fan-outs.
+    fn mock_target(id: &str, fail: bool) -> MockTarget {
+        let target = MockTarget::new(id, "webhook");
+        if fail { target.with_save_failures(usize::MAX) } else { target }
    }

    fn pipeline_with(targets: Vec<MockTarget>) -> AuditPipeline {
        let mut registry = AuditRegistry::new();
        for target in targets {
-            registry.add_target(target.id.to_string(), Box::new(target));
+            registry.add_target(target.target_id().to_string(), Box::new(target));
        }
        AuditPipeline::new(Arc::new(Mutex::new(registry)))
    }
@@ -658,7 +591,7 @@ mod tests {
    // dispatch must return Err rather than swallowing the failures as Ok.
    #[tokio::test]
    async fn dispatch_returns_err_when_all_targets_fail() {
-        let pipeline = pipeline_with(vec![MockTarget::new("a:webhook", true), MockTarget::new("b:webhook", true)]);
+        let pipeline = pipeline_with(vec![mock_target("a:webhook", true), mock_target("b:webhook", true)]);
        let result = pipeline.dispatch(entry()).await;
        assert!(matches!(result, Err(AuditError::Target(_))), "expected Err, got {result:?}");
    }
@@ -667,13 +600,13 @@ mod tests {
    // so dispatch reports success (degradation is logged, not propagated).
    #[tokio::test]
    async fn dispatch_returns_ok_on_partial_failure() {
-        let pipeline = pipeline_with(vec![MockTarget::new("ok:webhook", false), MockTarget::new("bad:webhook", true)]);
+        let pipeline = pipeline_with(vec![mock_target("ok:webhook", false), mock_target("bad:webhook", true)]);
        pipeline.dispatch(entry()).await.expect("partial success should return Ok");
    }

    #[tokio::test]
    async fn dispatch_returns_ok_when_all_targets_succeed() {
-        let pipeline = pipeline_with(vec![MockTarget::new("a:webhook", false), MockTarget::new("b:webhook", false)]);
+        let pipeline = pipeline_with(vec![mock_target("a:webhook", false), mock_target("b:webhook", false)]);
        pipeline.dispatch(entry()).await.expect("all-success should return Ok");
    }

@@ -686,9 +619,10 @@ mod tests {

    #[tokio::test]
    async fn health_probe_does_not_hold_the_registry_lock() {
-        let started = Arc::new(Notify::new());
        let release = Arc::new(Notify::new());
-        let pipeline = pipeline_with(vec![MockTarget::new("blocked", false).with_health_gate(started.clone(), release.clone())]);
+        let target = mock_target("blocked", false).with_health_gate(release.clone());
+        let started = target.health_started();
+        let pipeline = pipeline_with(vec![target]);
        let registry = Arc::clone(&pipeline.registry);
        let snapshot_task = tokio::spawn(async move { pipeline.snapshot_target_health().await });
        started.notified().await;
@@ -706,14 +640,14 @@ mod tests {
    // whole-batch loss instead of returning Ok.
    #[tokio::test]
    async fn dispatch_batch_returns_err_when_all_targets_fail() {
-        let pipeline = pipeline_with(vec![MockTarget::new("a:webhook", true)]);
+        let pipeline = pipeline_with(vec![mock_target("a:webhook", true)]);
        let result = pipeline.dispatch_batch(vec![entry(), entry()]).await;
        assert!(matches!(result, Err(AuditError::Target(_))), "expected Err, got {result:?}");
    }

    #[tokio::test]
    async fn dispatch_batch_returns_ok_when_all_targets_succeed() {
-        let pipeline = pipeline_with(vec![MockTarget::new("a:webhook", false), MockTarget::new("b:webhook", false)]);
+        let pipeline = pipeline_with(vec![mock_target("a:webhook", false), mock_target("b:webhook", false)]);
        pipeline
            .dispatch_batch(vec![entry(), entry()])
            .await
@@ -286,70 +286,10 @@ impl AuditRegistry {
 #[cfg(test)]
 mod tests {
    use super::AuditRegistry;
-    use crate::{AuditEntry, AuditError};
-    use rustfs_targets::arn::TargetID;
-    use rustfs_targets::store::{Key, Store};
-    use rustfs_targets::target::{ChannelTargetType, EntityTarget, QueuedPayload, QueuedPayloadMeta};
-    use rustfs_targets::{StoreError, Target, TargetError};
-    use std::sync::Arc;
-    use std::sync::atomic::{AtomicUsize, Ordering};
-
-    #[derive(Clone)]
-    struct CloseTestTarget {
-        id: TargetID,
-        close_calls: Arc<AtomicUsize>,
-        fail_on_close: bool,
-    }
-
-    impl CloseTestTarget {
-        fn new(id: TargetID, close_calls: Arc<AtomicUsize>, fail_on_close: bool) -> Self {
-            Self {
-                id,
-                close_calls,
-                fail_on_close,
-            }
-        }
-    }
-
-    #[async_trait::async_trait]
-    impl Target<AuditEntry> for CloseTestTarget {
-        fn id(&self) -> TargetID {
-            self.id.clone()
-        }
-
-        async fn is_active(&self) -> Result<bool, TargetError> {
-            Ok(true)
-        }
-
-        async fn save(&self, _event: Arc<EntityTarget<AuditEntry>>) -> Result<(), TargetError> {
-            Ok(())
-        }
-
-        async fn send_raw_from_store(&self, _key: Key, _body: Vec<u8>, _meta: QueuedPayloadMeta) -> Result<(), TargetError> {
-            Ok(())
-        }
-
-        async fn close(&self) -> Result<(), TargetError> {
-            self.close_calls.fetch_add(1, Ordering::SeqCst);
-            if self.fail_on_close {
-                Err(TargetError::Unknown("close failed".to_string()))
-            } else {
-                Ok(())
-            }
-        }
-
-        fn store(&self) -> Option<&(dyn Store<QueuedPayload, Error = StoreError, Key = Key> + Send + Sync)> {
-            None
-        }
-
-        fn clone_dyn(&self) -> Box<dyn Target<AuditEntry> + Send + Sync> {
-            Box::new(self.clone())
-        }
-
-        fn is_enabled(&self) -> bool {
-            true
-        }
-    }
+    use crate::AuditError;
+    use rustfs_targets::TargetError;
+    use rustfs_targets::target::ChannelTargetType;
+    use rustfs_targets::testkit::MockTarget;

    #[test]
    fn registry_registers_amqp_factory() {
@@ -361,23 +301,21 @@ mod tests {
    #[tokio::test]
    async fn close_all_returns_first_error_and_clears_targets() {
        let mut registry = AuditRegistry::new();
-        let ok_calls = Arc::new(AtomicUsize::new(0));
-        let fail_calls = Arc::new(AtomicUsize::new(0));
+        let ok = MockTarget::new("ok", "webhook");
+        let ok_observer = ok.clone();
+        let fail = MockTarget::new("fail", "webhook")
+            .with_close_failures(usize::MAX)
+            .with_close_failure_error(|| TargetError::Unknown("close failed".to_string()));
+        let fail_observer = fail.clone();

-        let ok_id = TargetID::new("ok".to_string(), "webhook".to_string());
-        let fail_id = TargetID::new("fail".to_string(), "webhook".to_string());
-
-        registry.add_target(ok_id.to_string(), Box::new(CloseTestTarget::new(ok_id, Arc::clone(&ok_calls), false)));
-        registry.add_target(
-            fail_id.to_string(),
-            Box::new(CloseTestTarget::new(fail_id, Arc::clone(&fail_calls), true)),
-        );
+        registry.add_target(ok.target_id().to_string(), Box::new(ok));
+        registry.add_target(fail.target_id().to_string(), Box::new(fail));

        let result = registry.close_all().await;

        assert!(matches!(result, Err(AuditError::Target(TargetError::Unknown(_)))));
-        assert_eq!(ok_calls.load(Ordering::SeqCst), 1);
-        assert_eq!(fail_calls.load(Ordering::SeqCst), 1);
+        assert_eq!(ok_observer.close_call_count(), 1);
+        assert_eq!(fail_observer.close_call_count(), 1);
        assert!(registry.list_targets().is_empty());
    }
 }
@@ -577,76 +577,17 @@ fn warn_audit_state(state: &str, reason: Option<&str>) {
 mod tests {
    use super::{AuditSystem, AuditSystemState};
    use crate::{AuditEntry, AuditError};
-    use async_trait::async_trait;
    use rustfs_targets::ReplayWorkerManager;
-    use rustfs_targets::arn::TargetID;
-    use rustfs_targets::store::{Key, Store};
-    use rustfs_targets::target::{EntityTarget, QueuedPayload, QueuedPayloadMeta};
-    use rustfs_targets::{StoreError, Target, TargetError};
+    use rustfs_targets::testkit::MockTarget;
    use std::collections::HashMap;
    use std::sync::Arc;
-    use std::sync::atomic::{AtomicUsize, Ordering};
    use tokio::sync::mpsc;

-    #[derive(Clone)]
-    struct TestTarget {
-        close_calls: Arc<AtomicUsize>,
-        id: TargetID,
-    }
-
-    impl TestTarget {
-        fn new(id: &str, name: &str) -> Self {
-            Self {
-                close_calls: Arc::new(AtomicUsize::new(0)),
-                id: TargetID::new(id.to_string(), name.to_string()),
-            }
-        }
-    }
-
-    #[async_trait]
-    impl<E> Target<E> for TestTarget
-    where
-        E: rustfs_targets::PluginEvent,
-    {
-        fn id(&self) -> TargetID {
-            self.id.clone()
-        }
-
-        async fn is_active(&self) -> Result<bool, TargetError> {
-            Ok(true)
-        }
-
-        async fn save(&self, _event: Arc<EntityTarget<E>>) -> Result<(), TargetError> {
-            Ok(())
-        }
-
-        async fn send_raw_from_store(&self, _key: Key, _body: Vec<u8>, _meta: QueuedPayloadMeta) -> Result<(), TargetError> {
-            Ok(())
-        }
-
-        async fn close(&self) -> Result<(), TargetError> {
-            self.close_calls.fetch_add(1, Ordering::SeqCst);
-            Ok(())
-        }
-
-        fn store(&self) -> Option<&(dyn Store<QueuedPayload, Error = StoreError, Key = Key> + Send + Sync)> {
-            None
-        }
-
-        fn clone_dyn(&self) -> Box<dyn Target<E> + Send + Sync> {
-            Box::new(self.clone())
-        }
-
-        fn is_enabled(&self) -> bool {
-            true
-        }
-    }
-
    #[tokio::test]
    async fn reload_with_empty_config_stops_existing_runtime() {
        let system = AuditSystem::new();
-        let target = TestTarget::new("primary", "webhook");
-        let close_calls = Arc::clone(&target.close_calls);
+        let target = MockTarget::new("primary", "webhook");
+        let observer = target.clone();

        {
            let mut registry = system.registry.lock().await;
@@ -671,7 +612,7 @@ mod tests {
        assert_eq!(system.get_state().await, AuditSystemState::Stopped);
        assert!(system.list_targets().await.is_empty());
        assert_eq!(system.runtime_status_snapshot().await, ReplayWorkerManager::new().snapshot(0));
-        assert_eq!(close_calls.load(Ordering::SeqCst), 1);
+        assert_eq!(observer.close_call_count(), 1);
        assert_eq!(*system.config.read().await, Some(rustfs_config::server_config::Config(HashMap::new())));
    }

@@ -693,7 +634,7 @@ mod tests {
        // Seed a target + replay worker so both critical sections touch real state.
        {
            let mut registry = system.registry.lock().await;
-            registry.add_target("primary:webhook".to_string(), Box::new(TestTarget::new("primary", "webhook")));
+            registry.add_target("primary:webhook".to_string(), Box::new(MockTarget::new("primary", "webhook")));
        }
        {
            let mut replay_workers = system.stream_cancellers.write().await;
@@ -793,8 +734,8 @@ mod tests {
    async fn commit_closes_old_targets_before_installing_new() {
        let system = AuditSystem::new();

-        let old = TestTarget::new("old", "webhook");
-        let old_close = Arc::clone(&old.close_calls);
+        let old = MockTarget::new("old", "webhook");
+        let old_observer = old.clone();
        {
            let mut registry = system.registry.lock().await;
            registry.add_target("old:webhook".to_string(), Box::new(old));
@@ -809,17 +750,17 @@ mod tests {
            *state = AuditSystemState::Running;
        }

-        let new = TestTarget::new("new", "webhook");
-        let new_close = Arc::clone(&new.close_calls);
+        let new = MockTarget::new("new", "webhook");
+        let new_observer = new.clone();
        system
            .commit_runtime_targets(vec![Box::new(new)], AuditSystemState::Running)
            .await
            .expect("commit should succeed");

        // Old target closed exactly once during the pre-install shutdown.
-        assert_eq!(old_close.load(Ordering::SeqCst), 1);
+        assert_eq!(old_observer.close_call_count(), 1);
        // New target installed and left open.
-        assert_eq!(new_close.load(Ordering::SeqCst), 0);
+        assert_eq!(new_observer.close_call_count(), 0);
        assert_eq!(system.list_targets().await, vec!["new:webhook".to_string()]);
        // Old replay worker stopped; the store-less new target adds none.
        assert_eq!(system.runtime_status_snapshot().await.replay_worker_count, 0);
@@ -12,136 +12,16 @@
 //  See the License for the specific language governing permissions and
 //  limitations under the License.

-use async_trait::async_trait;
 use rustfs_audit::{AuditEntry, AuditError, AuditPipeline, AuditRegistry, AuditRuntimeFacade, AuditRuntimeView};
-use rustfs_targets::arn::TargetID;
-use rustfs_targets::store::{Key, Store};
-use rustfs_targets::target::{EntityTarget, QueuedPayload, QueuedPayloadMeta};
-use rustfs_targets::{SharedTarget, StoreError, Target, TargetError};
-use serde::{Serialize, de::DeserializeOwned};
+use rustfs_targets::SharedTarget;
+use rustfs_targets::testkit::MockTarget;
 use std::sync::Arc;
-use std::sync::atomic::{AtomicUsize, Ordering};
 use tokio::sync::{Mutex, RwLock};

-#[derive(Clone)]
-struct TestTarget {
-    close_calls: Arc<AtomicUsize>,
-    id: TargetID,
-    init_calls: Arc<AtomicUsize>,
-}
-
-impl TestTarget {
-    fn new(id: &str, name: &str) -> Self {
-        Self {
-            close_calls: Arc::new(AtomicUsize::new(0)),
-            id: TargetID::new(id.to_string(), name.to_string()),
-            init_calls: Arc::new(AtomicUsize::new(0)),
-        }
-    }
-}
-
-#[async_trait]
-impl<E> Target<E> for TestTarget
-where
-    E: Send + Sync + 'static + Clone + Serialize + DeserializeOwned,
-{
-    fn id(&self) -> TargetID {
-        self.id.clone()
-    }
-
-    async fn is_active(&self) -> Result<bool, TargetError> {
-        Ok(true)
-    }
-
-    async fn save(&self, _event: Arc<EntityTarget<E>>) -> Result<(), TargetError> {
-        Ok(())
-    }
-
-    async fn send_raw_from_store(&self, _key: Key, _body: Vec<u8>, _meta: QueuedPayloadMeta) -> Result<(), TargetError> {
-        Ok(())
-    }
-
-    async fn close(&self) -> Result<(), TargetError> {
-        self.close_calls.fetch_add(1, Ordering::SeqCst);
-        Ok(())
-    }
-
-    fn store(&self) -> Option<&(dyn Store<QueuedPayload, Error = StoreError, Key = Key> + Send + Sync)> {
-        None
-    }
-
-    fn clone_dyn(&self) -> Box<dyn Target<E> + Send + Sync> {
-        Box::new(self.clone())
-    }
-
-    async fn init(&self) -> Result<(), TargetError> {
-        self.init_calls.fetch_add(1, Ordering::SeqCst);
-        Ok(())
-    }
-
-    fn is_enabled(&self) -> bool {
-        true
-    }
-}
-
-/// A target whose `save()` always fails, used to exercise the dispatch
+/// Builds a target whose `save()` always fails, used to exercise the dispatch
 /// failure-propagation paths.
-#[derive(Clone)]
-struct FailingTarget {
-    id: TargetID,
-    save_calls: Arc<AtomicUsize>,
-}
-
-impl FailingTarget {
-    fn new(id: &str, name: &str) -> Self {
-        Self {
-            id: TargetID::new(id.to_string(), name.to_string()),
-            save_calls: Arc::new(AtomicUsize::new(0)),
-        }
-    }
-}
-
-#[async_trait]
-impl<E> Target<E> for FailingTarget
-where
-    E: Send + Sync + 'static + Clone + Serialize + DeserializeOwned,
-{
-    fn id(&self) -> TargetID {
-        self.id.clone()
-    }
-
-    async fn is_active(&self) -> Result<bool, TargetError> {
-        Ok(true)
-    }
-
-    async fn save(&self, _event: Arc<EntityTarget<E>>) -> Result<(), TargetError> {
-        self.save_calls.fetch_add(1, Ordering::SeqCst);
-        Err(TargetError::Storage("disk full".to_string()))
-    }
-
-    async fn send_raw_from_store(&self, _key: Key, _body: Vec<u8>, _meta: QueuedPayloadMeta) -> Result<(), TargetError> {
-        Ok(())
-    }
-
-    async fn close(&self) -> Result<(), TargetError> {
-        Ok(())
-    }
-
-    fn store(&self) -> Option<&(dyn Store<QueuedPayload, Error = StoreError, Key = Key> + Send + Sync)> {
-        None
-    }
-
-    fn clone_dyn(&self) -> Box<dyn Target<E> + Send + Sync> {
-        Box::new(self.clone())
-    }
-
-    async fn init(&self) -> Result<(), TargetError> {
-        Ok(())
-    }
-
-    fn is_enabled(&self) -> bool {
-        true
-    }
+fn failing_target(id: &str, name: &str) -> MockTarget {
+    MockTarget::new(id, name).with_save_failures(usize::MAX)
 }

 fn pipeline_with_targets(targets: Vec<(&str, SharedTarget<AuditEntry>)>) -> AuditPipeline {
@@ -154,8 +34,8 @@ fn pipeline_with_targets(targets: Vec<(&str, SharedTarget<AuditEntry>)>) -> Audi

 #[tokio::test]
 async fn audit_pipeline_dispatch_propagates_total_failure() {
-    let failing = FailingTarget::new("primary", "webhook");
-    let save_calls = Arc::clone(&failing.save_calls);
+    let failing = failing_target("primary", "webhook");
+    let observer = failing.clone();
    let pipeline = pipeline_with_targets(vec![("primary:webhook", Arc::new(failing))]);

    let result = pipeline.dispatch(Arc::new(AuditEntry::default())).await;
@@ -164,13 +44,13 @@ async fn audit_pipeline_dispatch_propagates_total_failure() {
        matches!(result, Err(AuditError::Target(_))),
        "dispatch must surface an error when every target fails, got {result:?}"
    );
-    assert_eq!(save_calls.load(Ordering::SeqCst), 1, "the failing target should have been invoked");
+    assert_eq!(observer.save_call_count(), 1, "the failing target should have been invoked");
 }

 #[tokio::test]
 async fn audit_pipeline_dispatch_tolerates_partial_failure() {
-    let failing = FailingTarget::new("primary", "webhook");
-    let healthy = TestTarget::new("secondary", "webhook");
+    let failing = failing_target("primary", "webhook");
+    let healthy = MockTarget::new("secondary", "webhook");
    let pipeline = pipeline_with_targets(vec![
        ("primary:webhook", Arc::new(failing)),
        ("secondary:webhook", Arc::new(healthy)),
@@ -186,7 +66,7 @@ async fn audit_pipeline_dispatch_tolerates_partial_failure() {

 #[tokio::test]
 async fn audit_pipeline_dispatch_batch_propagates_total_failure() {
-    let failing = FailingTarget::new("primary", "webhook");
+    let failing = failing_target("primary", "webhook");
    let pipeline = pipeline_with_targets(vec![("primary:webhook", Arc::new(failing))]);

    let entries = vec![Arc::new(AuditEntry::default()), Arc::new(AuditEntry::default())];
@@ -200,8 +80,8 @@ async fn audit_pipeline_dispatch_batch_propagates_total_failure() {

 #[tokio::test]
 async fn audit_pipeline_dispatch_batch_tolerates_partial_failure() {
-    let failing = FailingTarget::new("primary", "webhook");
-    let healthy = TestTarget::new("secondary", "webhook");
+    let failing = failing_target("primary", "webhook");
+    let healthy = MockTarget::new("secondary", "webhook");
    let pipeline = pipeline_with_targets(vec![
        ("primary:webhook", Arc::new(failing)),
        ("secondary:webhook", Arc::new(healthy)),
@@ -266,9 +146,8 @@ async fn audit_runtime_facade_activates_empty_target_list() {
 async fn audit_runtime_view_upsert_and_remove_target() {
    let registry = Arc::new(Mutex::new(AuditRegistry::new()));
    let runtime_view = AuditRuntimeView::new(registry.clone());
-    let target = TestTarget::new("primary", "webhook");
-    let init_calls = Arc::clone(&target.init_calls);
-    let close_calls = Arc::clone(&target.close_calls);
+    let target = MockTarget::new("primary", "webhook");
+    let observer = target.clone();

    runtime_view
        .upsert_target("primary:webhook".to_string(), Box::new(target))
@@ -276,7 +155,7 @@ async fn audit_runtime_view_upsert_and_remove_target() {
        .expect("upsert should succeed");

    assert_eq!(runtime_view.list_targets().await, vec!["primary:webhook".to_string()]);
-    assert_eq!(init_calls.load(Ordering::SeqCst), 1);
+    assert_eq!(observer.init_call_count(), 1);

    runtime_view
        .remove_target("primary:webhook")
@@ -284,7 +163,7 @@ async fn audit_runtime_view_upsert_and_remove_target() {
        .expect("remove should succeed");

    assert!(runtime_view.list_targets().await.is_empty());
-    assert_eq!(close_calls.load(Ordering::SeqCst), 1);
+    assert_eq!(observer.close_call_count(), 1);
 }

 #[tokio::test]
@@ -292,7 +171,7 @@ async fn audit_runtime_facade_replace_targets_commits_runtime_state() {
    let registry = Arc::new(Mutex::new(AuditRegistry::new()));
    let replay_workers = Arc::new(RwLock::new(rustfs_targets::ReplayWorkerManager::new()));
    let facade = AuditRuntimeFacade::new(registry.clone(), replay_workers.clone());
-    let target = TestTarget::new("primary", "webhook");
+    let target = MockTarget::new("primary", "webhook");
    let activation = rustfs_targets::RuntimeActivation {
        replay_workers: rustfs_targets::ReplayWorkerManager::new(),
        targets: vec![Arc::new(target) as rustfs_targets::SharedTarget<rustfs_audit::AuditEntry>],
@@ -41,14 +41,22 @@ pub const XXHASH_64_NAME: &str = "xxhash64";
 pub const XXHASH_128_NAME: &str = "xxhash128";
 pub const MD5_NAME: &str = "md5";

-/// One of three deliberately separate checksum registries (backlog#1833):
-/// this enum owns the **streaming-hash algorithm registry**, including the
-/// RustFS extensions (sha512, xxhash3/64/128). The on-disk xl.meta bitset
-/// lives in `rustfs_rio::ChecksumType` (crates/rio/src/checksum.rs, varint
-/// bits are append-only), and the MinIO-port client keeps its own
-/// `ChecksumMode` (crates/ecstore/src/client/checksum.rs). When adding an
-/// algorithm, extend all three (or record why not) — they do not derive from
-/// each other.
+/// The canonical checksum-algorithm registry (backlog#1833, backlog#1844):
+/// this enum owns the streaming-hash implementations and, via the exhaustive
+/// per-algorithm metadata methods below, the wire names, header names, digest
+/// lengths, and checksum-type capabilities — including the RustFS extensions
+/// (sha512, xxhash3/64/128). The MinIO-port client's `ChecksumMode`
+/// (crates/s3-client/src/checksum.rs) delegates all per-algorithm dispatch
+/// here through its `algorithm()` bridge. The on-disk xl.meta bitset remains
+/// deliberately separate in `rustfs_rio::ChecksumType`
+/// (crates/rio/src/checksum.rs, varint bits are append-only), and rio also
+/// keeps its own hot-path hasher shells — equivalence with this crate's
+/// hashers is enforced by both test suites pinning the same official
+/// known-answer vectors (backlog#1844 PR3 verdict, recorded on
+/// `rustfs_rio::ChecksumType`). When adding an algorithm: add the variant
+/// here (the exhaustive matches force every metadata decision), bridge it in
+/// the client, and allocate an xl.meta bit + hasher + shared vector in rio
+/// (or record why not).
 #[derive(Debug, Clone, Copy, PartialEq, Eq, Default)]
 #[non_exhaustive]
 pub enum ChecksumAlgorithm {
@@ -120,6 +128,84 @@ impl ChecksumAlgorithm {
            Self::Xxhash128 => XXHASH_128_NAME,
        }
    }
+
+    // Per-algorithm wire metadata. These matches are deliberately exhaustive
+    // (no `_` arm): adding a ChecksumAlgorithm variant without deciding its
+    // name, header, digest length, and checksum-type support must fail to
+    // compile rather than silently inherit a default (backlog#1844).
+
+    /// The canonical `x-amz-checksum-algorithm` wire value (uppercase), as
+    /// carried in S3 requests/responses and stored checksum maps.
+    pub fn s3_algorithm_name(&self) -> &'static str {
+        match self {
+            Self::Crc32 => "CRC32",
+            Self::Crc32c => "CRC32C",
+            Self::Crc64Nvme => "CRC64NVME",
+            Self::Sha1 => "SHA1",
+            Self::Sha256 => "SHA256",
+            Self::Sha512 => "SHA512",
+            Self::Xxhash3 => "XXHASH3",
+            Self::Xxhash64 => "XXHASH64",
+            Self::Xxhash128 => "XXHASH128",
+        }
+    }
+
+    /// The `x-amz-checksum-*` HTTP header that carries this algorithm's
+    /// base64-encoded digest.
+    pub fn http_header_name(&self) -> &'static str {
+        match self {
+            Self::Crc32 => http::CRC_32_HEADER_NAME,
+            Self::Crc32c => http::CRC_32_C_HEADER_NAME,
+            Self::Crc64Nvme => http::CRC_64_NVME_HEADER_NAME,
+            Self::Sha1 => http::SHA_1_HEADER_NAME,
+            Self::Sha256 => http::SHA_256_HEADER_NAME,
+            Self::Sha512 => http::SHA_512_HEADER_NAME,
+            Self::Xxhash3 => http::XXHASH_3_HEADER_NAME,
+            Self::Xxhash64 => http::XXHASH_64_HEADER_NAME,
+            Self::Xxhash128 => http::XXHASH_128_HEADER_NAME,
+        }
+    }
+
+    /// Raw (unencoded) digest length in bytes.
+    pub fn raw_len(&self) -> usize {
+        match self {
+            Self::Crc32 | Self::Crc32c => 4,
+            Self::Crc64Nvme => 8,
+            Self::Sha1 => 20,
+            Self::Sha256 => 32,
+            Self::Sha512 => 64,
+            Self::Xxhash3 | Self::Xxhash64 => 8,
+            Self::Xxhash128 => 16,
+        }
+    }
+
+    /// Whether the algorithm supports the S3 COMPOSITE multipart checksum
+    /// type. Per the AWS registry, every algorithm does except CRC64NVME,
+    /// which is FULL_OBJECT-only.
+    pub fn supports_composite(&self) -> bool {
+        match self {
+            Self::Crc64Nvme => false,
+            Self::Crc32
+            | Self::Crc32c
+            | Self::Sha1
+            | Self::Sha256
+            | Self::Sha512
+            | Self::Xxhash3
+            | Self::Xxhash64
+            | Self::Xxhash128 => true,
+        }
+    }
+
+    /// Whether the algorithm supports the S3 FULL_OBJECT checksum type, i.e.
+    /// part digests can be linearly combined into the whole-object digest.
+    /// Only the CRC family has this property; the hash algorithms are
+    /// COMPOSITE-only.
+    pub fn supports_full_object(&self) -> bool {
+        match self {
+            Self::Crc32 | Self::Crc32c | Self::Crc64Nvme => true,
+            Self::Sha1 | Self::Sha256 | Self::Sha512 | Self::Xxhash3 | Self::Xxhash64 | Self::Xxhash128 => false,
+        }
+    }
 }

 pub trait Checksum: Send + Sync {
@@ -731,6 +817,77 @@ mod tests {
        assert_eq!(&raw[..], reference.digest128().to_be_bytes().as_slice());
    }

+    #[test]
+    fn test_algorithm_metadata_is_consistent_for_every_variant() {
+        use crate::Checksum;
+
+        // Cross-checks the per-algorithm metadata methods against the hasher
+        // implementations themselves, so the registry cannot drift from the
+        // code that computes digests (backlog#1844). The list must cover every
+        // variant; the metadata methods use exhaustive matches, so a new
+        // variant that is missing here still fails to compile there first.
+        let all = [
+            ChecksumAlgorithm::Crc32,
+            ChecksumAlgorithm::Crc32c,
+            ChecksumAlgorithm::Crc64Nvme,
+            ChecksumAlgorithm::Sha1,
+            ChecksumAlgorithm::Sha256,
+            ChecksumAlgorithm::Sha512,
+            ChecksumAlgorithm::Xxhash3,
+            ChecksumAlgorithm::Xxhash64,
+            ChecksumAlgorithm::Xxhash128,
+        ];
+
+        for algorithm in all {
+            // Digest length must match what the hasher actually produces.
+            let mut hasher = algorithm.into_impl();
+            hasher.update(b"metadata consistency probe");
+            assert_eq!(
+                algorithm.raw_len(),
+                Checksum::size(&*algorithm.into_impl()) as usize,
+                "{algorithm:?} raw_len() != hasher size()"
+            );
+            assert_eq!(hasher.finalize().len(), algorithm.raw_len(), "{algorithm:?} finalize length != raw_len()");
+
+            // Header name must match the hasher's own header binding.
+            assert_eq!(
+                algorithm.http_header_name(),
+                algorithm.into_impl().header_name(),
+                "{algorithm:?} http_header_name() != HttpChecksum::header_name()"
+            );
+            assert_eq!(
+                algorithm.http_header_name(),
+                format!("x-amz-checksum-{}", algorithm.as_str()),
+                "{algorithm:?} header must be x-amz-checksum-<name>"
+            );
+
+            // The uppercase wire name and the lowercase parse name must be the
+            // same word, and the wire name must parse back to the variant.
+            assert!(
+                algorithm.s3_algorithm_name().eq_ignore_ascii_case(algorithm.as_str()),
+                "{algorithm:?} s3_algorithm_name() and as_str() diverge"
+            );
+            assert_eq!(algorithm.s3_algorithm_name().parse::<ChecksumAlgorithm>().unwrap(), algorithm);
+        }
+
+        // AWS checksum-type support table: CRC64NVME is FULL_OBJECT-only, the
+        // CRC family supports FULL_OBJECT, everything else is COMPOSITE-only.
+        for algorithm in all {
+            let composite = algorithm.supports_composite();
+            let full_object = algorithm.supports_full_object();
+            assert!(composite || full_object, "{algorithm:?} supports no checksum type at all");
+            match algorithm {
+                ChecksumAlgorithm::Crc32 | ChecksumAlgorithm::Crc32c => {
+                    assert!(composite && full_object, "{algorithm:?} must support both checksum types")
+                }
+                ChecksumAlgorithm::Crc64Nvme => {
+                    assert!(!composite && full_object, "CRC64NVME must be FULL_OBJECT-only")
+                }
+                _ => assert!(composite && !full_object, "{algorithm:?} must be COMPOSITE-only"),
+            }
+        }
+    }
+
    #[test]
    fn test_xxhash64_matches_direct_computation_big_endian_seed0() {
        use crate::Xxhash64;
@@ -38,16 +38,9 @@ hotpath.workspace = true
 tokio = { workspace = true, features = ["fs", "rt-multi-thread"] }
 tonic = { workspace = true, features = ["gzip", "deflate"] }
 uuid = { workspace = true, features = ["v4", "fast-rng", "macro-diagnostics"] }
-chrono = { workspace = true, features = ["serde"] }
-jiff = { workspace = true, features = ["serde"] }
 metrics = { workspace = true }
-serde = { workspace = true, features = ["derive"] }
 smallvec = { workspace = true }
-rmp-serde = { workspace = true }
 tracing = { workspace = true }

-[dev-dependencies]
-serde_json = { workspace = true }
-
 [lib]
 doctest = false
@@ -12,7 +12,6 @@
 // See the License for the specific language governing permissions and
 // limitations under the License.

-use chrono::{DateTime, Utc};
 use std::collections::HashMap;
 use std::sync::LazyLock;
 use std::sync::atomic::{AtomicU64, Ordering};
@@ -27,8 +26,6 @@ pub static GLOBAL_CONN_MAP: LazyLock<RwLock<HashMap<String, Channel>>> = LazyLoc
 pub static GLOBAL_ROOT_CERT: LazyLock<RwLock<Option<Vec<u8>>>> = LazyLock::new(|| RwLock::new(None));
 pub static GLOBAL_MTLS_IDENTITY: LazyLock<RwLock<Option<MtlsIdentityPem>>> = LazyLock::new(|| RwLock::new(None));
 pub static GLOBAL_OUTBOUND_TLS_GENERATION: LazyLock<AtomicU64> = LazyLock::new(|| AtomicU64::new(0));
-/// Global initialization time of the RustFS node.
-pub static GLOBAL_INIT_TIME: LazyLock<RwLock<Option<DateTime<Utc>>>> = LazyLock::new(|| RwLock::new(None));

 /// Log level to use when reporting cached gRPC connection eviction.
 #[derive(Clone, Copy, Debug, Eq, PartialEq)]
@@ -63,20 +60,6 @@ pub fn try_get_global_local_node_name() -> Option<String> {
        .filter(|name| !name.is_empty())
 }

-/// Set the global RustFS initialization time to the current UTC time.
-pub async fn set_global_init_time_now() {
-    let now = Utc::now();
-    *GLOBAL_INIT_TIME.write().await = Some(now);
-}
-
-/// Get the global RustFS initialization time.
-///
-/// # Returns
-/// * `Option<DateTime<Utc>>` - The initialization time if set.
-pub async fn get_global_init_time() -> Option<DateTime<Utc>> {
-    *GLOBAL_INIT_TIME.read().await
-}
-
 /// Set the global RustFS address used for gRPC connections.
 ///
 /// # Arguments
@@ -14,9 +14,6 @@

 // pub mod error;
 pub mod globals;
-pub mod heal_channel;
-pub mod last_minute;
-pub mod metrics;
 pub mod mrf_channel;
 mod readiness;
 pub mod table_catalog;
@@ -178,6 +178,76 @@ pub trait WorkloadAdmissionSnapshotProvider {
    fn workload_admission_snapshot(&self) -> WorkloadAdmissionRegistrySnapshot;
 }

+/// Foreground workload pressure observed against a configured utilization threshold.
+#[derive(Debug, Clone, Copy, PartialEq, Eq)]
+pub struct ForegroundPressure {
+    /// Foreground workload class whose utilization reached its threshold.
+    pub class: WorkloadClass,
+    /// Observed utilization percentage for the class.
+    pub usage_pct: usize,
+    /// Configured threshold percentage that the observed utilization reached.
+    pub threshold_pct: usize,
+}
+
+impl ForegroundPressure {
+    /// Return a stable reason label for logs and metrics.
+    pub const fn reason(self) -> &'static str {
+        match self.class {
+            WorkloadClass::ForegroundRead => "foreground_read_pressure",
+            WorkloadClass::ForegroundWrite => "foreground_write_pressure",
+            _ => "foreground_pressure",
+        }
+    }
+}
+
+/// Return the strongest foreground pressure in `snapshot`, if any.
+///
+/// A zero threshold disables its class. `Saturated` counts as full utilization
+/// regardless of the reported limit; otherwise a class contributes only when it
+/// reports a non-zero limit, with a missing active count read as zero. When both
+/// classes are above their threshold the higher utilization wins.
+///
+/// Callers own the enable switch: this function evaluates thresholds only.
+pub fn foreground_pressure(
+    snapshot: &WorkloadAdmissionRegistrySnapshot,
+    read_threshold_pct: usize,
+    write_threshold_pct: usize,
+) -> Option<ForegroundPressure> {
+    [
+        (WorkloadClass::ForegroundRead, read_threshold_pct),
+        (WorkloadClass::ForegroundWrite, write_threshold_pct),
+    ]
+    .into_iter()
+    .filter_map(|(class, threshold_pct)| {
+        if threshold_pct == 0 {
+            return None;
+        }
+
+        let entry = snapshot.get(class)?;
+        let usage_pct = if matches!(entry.state, AdmissionState::Saturated) {
+            100
+        } else {
+            let limit = entry.limit?;
+            if limit == 0 {
+                return None;
+            }
+            entry
+                .active
+                .unwrap_or(0)
+                .saturating_mul(100)
+                .checked_div(limit)
+                .unwrap_or(100)
+        };
+
+        (usage_pct >= threshold_pct).then_some(ForegroundPressure {
+            class,
+            usage_pct,
+            threshold_pct,
+        })
+    })
+    .max_by_key(|pressure| pressure.usage_pct)
+}
+
 #[cfg(test)]
 mod tests {
    use super::*;
@@ -314,4 +384,205 @@ mod tests {

        assert!(err.to_string().contains("unexpected"));
    }
+
+    fn counted(
+        class: WorkloadClass,
+        state: AdmissionState,
+        active: Option<usize>,
+        limit: Option<usize>,
+    ) -> WorkloadAdmissionSnapshot {
+        WorkloadAdmissionSnapshot::new(class, state).with_counts(active, None, limit)
+    }
+
+    fn registry(entries: Vec<WorkloadAdmissionSnapshot>) -> WorkloadAdmissionRegistrySnapshot {
+        WorkloadAdmissionRegistrySnapshot::new(entries)
+    }
+
+    #[test]
+    fn foreground_pressure_reason_labels_cover_non_foreground_classes() {
+        let read = ForegroundPressure {
+            class: WorkloadClass::ForegroundRead,
+            usage_pct: 90,
+            threshold_pct: 80,
+        };
+        let write = ForegroundPressure {
+            class: WorkloadClass::ForegroundWrite,
+            usage_pct: 90,
+            threshold_pct: 80,
+        };
+        let repair = ForegroundPressure {
+            class: WorkloadClass::Repair,
+            usage_pct: 90,
+            threshold_pct: 80,
+        };
+
+        assert_eq!(read.reason(), "foreground_read_pressure");
+        assert_eq!(write.reason(), "foreground_write_pressure");
+        assert_eq!(repair.reason(), "foreground_pressure");
+    }
+
+    #[test]
+    fn foreground_pressure_is_disabled_when_both_thresholds_are_zero() {
+        let snapshot = registry(vec![
+            counted(WorkloadClass::ForegroundRead, AdmissionState::Saturated, Some(8), Some(8)),
+            counted(WorkloadClass::ForegroundWrite, AdmissionState::Saturated, Some(8), Some(8)),
+        ]);
+
+        assert_eq!(foreground_pressure(&snapshot, 0, 0), None);
+    }
+
+    #[test]
+    fn foreground_pressure_skips_only_the_class_whose_threshold_is_zero() {
+        let snapshot = registry(vec![
+            counted(WorkloadClass::ForegroundRead, AdmissionState::Open, Some(10), Some(10)),
+            counted(WorkloadClass::ForegroundWrite, AdmissionState::Open, Some(9), Some(10)),
+        ]);
+
+        assert_eq!(
+            foreground_pressure(&snapshot, 0, 80),
+            Some(ForegroundPressure {
+                class: WorkloadClass::ForegroundWrite,
+                usage_pct: 90,
+                threshold_pct: 80,
+            })
+        );
+        assert_eq!(
+            foreground_pressure(&snapshot, 80, 0),
+            Some(ForegroundPressure {
+                class: WorkloadClass::ForegroundRead,
+                usage_pct: 100,
+                threshold_pct: 80,
+            })
+        );
+    }
+
+    #[test]
+    fn foreground_pressure_ignores_missing_entries() {
+        let snapshot = registry(vec![counted(WorkloadClass::Scanner, AdmissionState::Saturated, Some(8), Some(8))]);
+
+        assert_eq!(foreground_pressure(&snapshot, 1, 1), None);
+    }
+
+    #[test]
+    fn foreground_pressure_ignores_missing_and_zero_limits() {
+        let missing_limit = registry(vec![counted(
+            WorkloadClass::ForegroundRead,
+            AdmissionState::Throttled,
+            Some(8),
+            None,
+        )]);
+        let zero_limit = registry(vec![counted(
+            WorkloadClass::ForegroundWrite,
+            AdmissionState::Throttled,
+            Some(8),
+            Some(0),
+        )]);
+
+        assert_eq!(foreground_pressure(&missing_limit, 1, 1), None);
+        assert_eq!(foreground_pressure(&zero_limit, 1, 1), None);
+    }
+
+    #[test]
+    fn foreground_pressure_treats_saturated_as_full_without_reading_limit() {
+        let snapshot = registry(vec![
+            counted(WorkloadClass::ForegroundRead, AdmissionState::Saturated, None, None),
+            counted(WorkloadClass::ForegroundWrite, AdmissionState::Saturated, Some(0), Some(0)),
+        ]);
+
+        assert_eq!(
+            foreground_pressure(&snapshot, 100, 0),
+            Some(ForegroundPressure {
+                class: WorkloadClass::ForegroundRead,
+                usage_pct: 100,
+                threshold_pct: 100,
+            })
+        );
+        assert_eq!(
+            foreground_pressure(&snapshot, 0, 100),
+            Some(ForegroundPressure {
+                class: WorkloadClass::ForegroundWrite,
+                usage_pct: 100,
+                threshold_pct: 100,
+            })
+        );
+    }
+
+    #[test]
+    fn foreground_pressure_reads_missing_active_as_zero() {
+        let snapshot = registry(vec![counted(WorkloadClass::ForegroundRead, AdmissionState::Open, None, Some(8))]);
+
+        assert_eq!(foreground_pressure(&snapshot, 1, 1), None);
+    }
+
+    #[test]
+    fn foreground_pressure_returns_the_higher_utilization_when_both_classes_exceed() {
+        let read_higher = registry(vec![
+            counted(WorkloadClass::ForegroundRead, AdmissionState::Open, Some(19), Some(20)),
+            counted(WorkloadClass::ForegroundWrite, AdmissionState::Open, Some(17), Some(20)),
+        ]);
+        let write_higher = registry(vec![
+            counted(WorkloadClass::ForegroundRead, AdmissionState::Open, Some(17), Some(20)),
+            counted(WorkloadClass::ForegroundWrite, AdmissionState::Open, Some(19), Some(20)),
+        ]);
+
+        assert_eq!(
+            foreground_pressure(&read_higher, 80, 80),
+            Some(ForegroundPressure {
+                class: WorkloadClass::ForegroundRead,
+                usage_pct: 95,
+                threshold_pct: 80,
+            })
+        );
+        assert_eq!(
+            foreground_pressure(&write_higher, 80, 80),
+            Some(ForegroundPressure {
+                class: WorkloadClass::ForegroundWrite,
+                usage_pct: 95,
+                threshold_pct: 80,
+            })
+        );
+    }
+
+    #[test]
+    fn foreground_pressure_breaks_utilization_ties_toward_the_write_class() {
+        let snapshot = registry(vec![
+            counted(WorkloadClass::ForegroundRead, AdmissionState::Open, Some(18), Some(20)),
+            counted(WorkloadClass::ForegroundWrite, AdmissionState::Open, Some(18), Some(20)),
+        ]);
+
+        assert_eq!(
+            foreground_pressure(&snapshot, 80, 80),
+            Some(ForegroundPressure {
+                class: WorkloadClass::ForegroundWrite,
+                usage_pct: 90,
+                threshold_pct: 80,
+            })
+        );
+    }
+
+    #[test]
+    fn foreground_pressure_triggers_exactly_at_the_threshold_and_not_below() {
+        let at_threshold = registry(vec![counted(
+            WorkloadClass::ForegroundRead,
+            AdmissionState::Open,
+            Some(8),
+            Some(10),
+        )]);
+        let below_threshold = registry(vec![counted(
+            WorkloadClass::ForegroundRead,
+            AdmissionState::Open,
+            Some(7),
+            Some(10),
+        )]);
+
+        assert_eq!(
+            foreground_pressure(&at_threshold, 80, 80),
+            Some(ForegroundPressure {
+                class: WorkloadClass::ForegroundRead,
+                usage_pct: 80,
+                threshold_pct: 80,
+            })
+        );
+        assert_eq!(foreground_pressure(&below_threshold, 80, 80), None);
+    }
 }
@@ -150,9 +150,10 @@ Drive timeout health-action policy:
 Drive timeout profile preset:
 - `RUSTFS_DRIVE_TIMEOUT_PROFILE`
  - `default` (default): keep current timeout defaults.
-  - `high_latency`: use 60s default timeout for scanner-sensitive operations when no per-operation timeout override is set (`read_metadata`, `disk_info`, `list_dir`, `walk_dir`, `walk_dir_stall`).
+  - `high_latency`: use 60s default timeout for scanner-sensitive operations when no operation-specific override is set (`read_metadata`, `disk_info`, `list_dir`, `walk_dir`, `walk_dir_stall`, and object-capacity scan base/maximum budgets).
 - Precedence:
  - Explicit per-operation timeout env (`RUSTFS_DRIVE_*_TIMEOUT_SECS`) takes highest precedence.
+  - Explicit object-capacity timeout env (`RUSTFS_CAPACITY_STAT_TIMEOUT`, `RUSTFS_CAPACITY_MAX_TIMEOUT`) takes precedence for capacity scans.
  - Then `RUSTFS_DRIVE_MAX_TIMEOUT_DURATION` legacy fallback.
  - Then the profile-derived default (`default` or `high_latency`).

@@ -59,20 +59,20 @@ pub const ENV_CAPACITY_MAX_TIMEOUT: &str = "RUSTFS_CAPACITY_MAX_TIMEOUT";
 // ============================================================================

 /// Scheduled update interval in seconds
-/// Default: 120 seconds (2 minutes)
-pub const DEFAULT_SCHEDULED_UPDATE_INTERVAL_SECS: u64 = 120;
+/// Default: 600 seconds (10 minutes)
+pub const DEFAULT_SCHEDULED_UPDATE_INTERVAL_SECS: u64 = 600;

 /// Write trigger delay in seconds
-/// Default: 5 seconds
-pub const DEFAULT_WRITE_TRIGGER_DELAY_SECS: u64 = 5;
+/// Default: 30 seconds
+pub const DEFAULT_WRITE_TRIGGER_DELAY_SECS: u64 = 30;

 /// Write frequency threshold (writes per minute)
-/// Default: 5 writes/minute
-pub const DEFAULT_WRITE_FREQUENCY_THRESHOLD: usize = 5;
+/// Default: 20 writes/minute
+pub const DEFAULT_WRITE_FREQUENCY_THRESHOLD: usize = 20;

 /// Fast update threshold in seconds
-/// Default: 30 seconds
-pub const DEFAULT_FAST_UPDATE_THRESHOLD_SECS: u64 = 30;
+/// Default: 120 seconds
+pub const DEFAULT_FAST_UPDATE_THRESHOLD_SECS: u64 = 120;

 /// Maximum files threshold for sampling
 /// Default: 200,000 files
@@ -129,4 +129,16 @@ mod tests {
        assert_eq!(ENV_CAPACITY_MIN_TIMEOUT, "RUSTFS_CAPACITY_MIN_TIMEOUT");
        assert_eq!(ENV_CAPACITY_MAX_TIMEOUT, "RUSTFS_CAPACITY_MAX_TIMEOUT");
    }
+
+    #[test]
+    fn test_capacity_default_values() {
+        assert_eq!(DEFAULT_SCHEDULED_UPDATE_INTERVAL_SECS, 600);
+        assert_eq!(DEFAULT_WRITE_TRIGGER_DELAY_SECS, 30);
+        assert_eq!(DEFAULT_WRITE_FREQUENCY_THRESHOLD, 20);
+        assert_eq!(DEFAULT_FAST_UPDATE_THRESHOLD_SECS, 120);
+        assert_eq!(DEFAULT_MAX_FILES_THRESHOLD, 200_000);
+        assert_eq!(DEFAULT_STAT_TIMEOUT_SECS, 3);
+        assert_eq!(DEFAULT_SAMPLE_RATE, 200);
+        assert_eq!(DEFAULT_CAPACITY_METRICS_INTERVAL_SECS, 600);
+    }
 }
@@ -40,6 +40,14 @@ pub const HEALTH_OBJECT_PROGRESS_LOCK_MARGIN_MS: u64 = 5_000;
 pub const ENV_HEALTH_CLUSTER_TIMEOUT_MS: &str = "RUSTFS_HEALTH_CLUSTER_TIMEOUT_MS";
 pub const DEFAULT_HEALTH_CLUSTER_TIMEOUT_MS: u64 = 2000;

+/// Timeout for one remote lock-client online check used by readiness (milliseconds).
+///
+/// This is intentionally shorter than the generic lock RPC timeout so
+/// `/health/ready` can report degradation instead of riding a dead peer's
+/// connect or HTTP/2 keepalive budget.
+pub const ENV_HEALTH_LOCK_ONLINE_TIMEOUT_MS: &str = "RUSTFS_HEALTH_LOCK_ONLINE_TIMEOUT_MS";
+pub const DEFAULT_HEALTH_LOCK_ONLINE_TIMEOUT_MS: u64 = 1000;
+
 /// Maximum time to wait for local node runtime readiness (storage / IAM / lock
 /// quorum) during startup before failing fast (seconds).
 ///
@@ -168,6 +168,35 @@ pub const DEFAULT_DATA_MOVEMENT_PART_CHECKSUMS_FLEET_CONFIRMED: bool = false;
 const _: () = assert!(!DEFAULT_DATA_MOVEMENT_PART_CHECKSUMS_WRITE);
 const _: () = assert!(!DEFAULT_DATA_MOVEMENT_PART_CHECKSUMS_FLEET_CONFIRMED);

+/// Request writing pool metadata version 2.
+///
+/// This remains ineffective until [`ENV_POOL_META_V2_FLEET_CONFIRMED`] is also enabled.
+pub const ENV_POOL_META_V2_WRITE: &str = "RUSTFS_POOL_META_V2_WRITE";
+pub const DEFAULT_POOL_META_V2_WRITE: bool = false;
+
+/// Operator-attested confirmation that every pool metadata reader and writer understands version 2.
+pub const ENV_POOL_META_V2_FLEET_CONFIRMED: &str = "RUSTFS_POOL_META_V2_FLEET_CONFIRMED";
+pub const DEFAULT_POOL_META_V2_FLEET_CONFIRMED: bool = false;
+
+const _: () = assert!(!DEFAULT_POOL_META_V2_WRITE);
+const _: () = assert!(!DEFAULT_POOL_META_V2_FLEET_CONFIRMED);
+
+/// Request writing pool metadata version 3 with durable generations.
+///
+/// Existing deployments remain on their observed version until
+/// [`ENV_POOL_META_V3_FLEET_CONFIRMED`] is also enabled. Fresh deployments may
+/// initialize directly at version 3 because they have no legacy readers.
+pub const ENV_POOL_META_V3_WRITE: &str = "RUSTFS_POOL_META_V3_WRITE";
+pub const DEFAULT_POOL_META_V3_WRITE: bool = false;
+
+/// Operator-attested confirmation that every pool metadata reader and writer
+/// understands the version 3 generation and recovery protocol.
+pub const ENV_POOL_META_V3_FLEET_CONFIRMED: &str = "RUSTFS_POOL_META_V3_FLEET_CONFIRMED";
+pub const DEFAULT_POOL_META_V3_FLEET_CONFIRMED: bool = false;
+
+const _: () = assert!(!DEFAULT_POOL_META_V3_WRITE);
+const _: () = assert!(!DEFAULT_POOL_META_V3_FLEET_CONFIRMED);
+
 // =============================================================================
 // Concurrent Request Fix - Timeout and Backpressure Configuration
 // =============================================================================
@@ -259,6 +288,49 @@ pub const DEFAULT_PUT_FOREGROUND_ADMISSION_WAIT_TIMEOUT_MS: u64 = 0;

 const _: () = assert!(!DEFAULT_PUT_FOREGROUND_ADMISSION_ENABLE);

+/// Enable automatic foreground admission for large or unknown-size PutObject requests.
+///
+/// Unlike the strict experimental gate above, this default-on path only applies
+/// to requests that are large enough to create sustained erasure/RPC pressure.
+/// Small PUTs continue on the legacy path unless the strict gate is explicitly
+/// enabled.
+pub const ENV_PUT_LARGE_FOREGROUND_ADMISSION_ENABLE: &str = "RUSTFS_PUT_LARGE_FOREGROUND_ADMISSION_ENABLE";
+pub const DEFAULT_PUT_LARGE_FOREGROUND_ADMISSION_ENABLE: bool = true;
+
+/// Maximum automatic foreground write requests admitted concurrently per process.
+///
+/// `0` derives a conservative default from the local disk-read scheduler cap,
+/// currently clamped to protect the commit path without making ordinary high
+/// throughput uploads single-file.
+pub const ENV_PUT_LARGE_FOREGROUND_ADMISSION_LIMIT: &str = "RUSTFS_PUT_LARGE_FOREGROUND_ADMISSION_LIMIT";
+pub const DEFAULT_PUT_LARGE_FOREGROUND_ADMISSION_LIMIT: usize = 0;
+
+/// Minimum direct PutObject size that enters automatic foreground write admission.
+///
+/// Requests with an unknown size are treated as large because the write pressure
+/// cannot be bounded from headers.
+pub const ENV_PUT_LARGE_FOREGROUND_ADMISSION_MIN_SIZE_BYTES: &str = "RUSTFS_PUT_LARGE_FOREGROUND_ADMISSION_MIN_SIZE_BYTES";
+pub const DEFAULT_PUT_LARGE_FOREGROUND_ADMISSION_MIN_SIZE_BYTES: usize = 32 * 1024 * 1024;
+
+/// Minimum UploadPart size that enters automatic foreground write admission.
+///
+/// Multipart pressure is often many moderate-sized parts rather than one very
+/// large request. The default gates every multipart part through the same permit
+/// pool as large/unknown-size PutObject while keeping small direct PUTs on the
+/// legacy path.
+pub const ENV_PUT_MULTIPART_FOREGROUND_ADMISSION_MIN_SIZE_BYTES: &str =
+    "RUSTFS_PUT_MULTIPART_FOREGROUND_ADMISSION_MIN_SIZE_BYTES";
+pub const DEFAULT_PUT_MULTIPART_FOREGROUND_ADMISSION_MIN_SIZE_BYTES: usize = 0;
+
+/// Time in milliseconds an automatic foreground write waits for a permit.
+///
+/// A short wait smooths transient bursts while still returning S3
+/// `SlowDown`/503 before body ingest when the node is already saturated.
+pub const ENV_PUT_LARGE_FOREGROUND_ADMISSION_WAIT_TIMEOUT_MS: &str = "RUSTFS_PUT_LARGE_FOREGROUND_ADMISSION_WAIT_TIMEOUT_MS";
+pub const DEFAULT_PUT_LARGE_FOREGROUND_ADMISSION_WAIT_TIMEOUT_MS: u64 = 250;
+
+const _: () = assert!(DEFAULT_PUT_LARGE_FOREGROUND_ADMISSION_ENABLE);
+
 /// Environment variable for minimum GetObject timeout in seconds.
 ///
 /// When dynamic timeout calculation is enabled, this is the minimum timeout
@@ -736,4 +808,16 @@ mod remote_version_state_tests {
            "RUSTFS_OBJECT_TRANSACTION_FENCING_FLEET_CONFIRMED"
        );
    }
+
+    #[test]
+    fn pool_meta_v2_gate_uses_stable_environment_names() {
+        assert_eq!(super::ENV_POOL_META_V2_WRITE, "RUSTFS_POOL_META_V2_WRITE");
+        assert_eq!(super::ENV_POOL_META_V2_FLEET_CONFIRMED, "RUSTFS_POOL_META_V2_FLEET_CONFIRMED");
+    }
+
+    #[test]
+    fn pool_meta_v3_gate_uses_stable_environment_names() {
+        assert_eq!(super::ENV_POOL_META_V3_WRITE, "RUSTFS_POOL_META_V3_WRITE");
+        assert_eq!(super::ENV_POOL_META_V3_FLEET_CONFIRMED, "RUSTFS_POOL_META_V3_FLEET_CONFIRMED");
+    }
 }
@@ -103,7 +103,7 @@ pub const ENV_ALLOCATOR_RECLAIM_ENABLED: &str = "RUSTFS_ALLOCATOR_RECLAIM_ENABLE
 pub const ENV_ALLOCATOR_RECLAIM_INTERVAL_SECS: &str = "RUSTFS_ALLOCATOR_RECLAIM_INTERVAL_SECS";
 pub const ENV_ALLOCATOR_RECLAIM_FORCE: &str = "RUSTFS_ALLOCATOR_RECLAIM_FORCE";
 pub const ENV_ALLOCATOR_RECLAIM_IDLE_INTERVALS: &str = "RUSTFS_ALLOCATOR_RECLAIM_IDLE_INTERVALS";
-pub const DEFAULT_ALLOCATOR_RECLAIM_ENABLED: bool = false;
+pub const DEFAULT_ALLOCATOR_RECLAIM_ENABLED: bool = true;
 pub const DEFAULT_ALLOCATOR_RECLAIM_INTERVAL_SECS: u64 = 30;
 pub const DEFAULT_ALLOCATOR_RECLAIM_FORCE: bool = true;
 pub const DEFAULT_ALLOCATOR_RECLAIM_IDLE_INTERVALS: u64 = 3;
@@ -198,11 +198,11 @@ pub const ENV_SCANNER_IDLE_MODE: &str = "RUSTFS_SCANNER_IDLE_MODE";
 /// Environment variable that controls scanner cache save timeout in seconds.
 /// The scanner enforces a minimum value of `1`.
 /// - Unit: seconds (u64).
-/// - Example: `export RUSTFS_SCANNER_CACHE_SAVE_TIMEOUT_SECS=30`
+/// - Example: `export RUSTFS_SCANNER_CACHE_SAVE_TIMEOUT_SECS=14`
 pub const ENV_SCANNER_CACHE_SAVE_TIMEOUT_SECS: &str = "RUSTFS_SCANNER_CACHE_SAVE_TIMEOUT_SECS";

 /// Default scanner cache save timeout in seconds.
-pub const DEFAULT_SCANNER_CACHE_SAVE_TIMEOUT_SECS: u64 = 30;
+pub const DEFAULT_SCANNER_CACHE_SAVE_TIMEOUT_SECS: u64 = 14;

 /// Environment variable that caps concurrent scanner set tasks.
 /// A value of `0` keeps the existing topology-based concurrency.
@@ -573,6 +573,10 @@ pub struct DataUsageInfo {
    /// cycle (or retained a compatible last-known-good cache).
    #[serde(default)]
    pub usage_snapshot_partial: bool,
+    /// Durable marker for a first-start bootstrap that has not produced an
+    /// authoritative usage snapshot yet.
+    #[serde(default, skip_serializing_if = "std::ops::Not::not")]
+    pub usage_snapshot_bootstrap_pending: bool,
    /// Deprecated kept here for backward compatibility reasons
    pub bucket_sizes: HashMap<String, u64>,
    /// Per-disk snapshot information when available
@@ -1962,7 +1966,8 @@ impl DataUsageInfo {

    /// Whether this snapshot authoritatively covers every reported bucket.
    pub fn is_complete_bucket_usage_snapshot(&self) -> bool {
-        self.usage_snapshot_complete
+        !self.usage_snapshot_bootstrap_pending
+            && self.usage_snapshot_complete
            && self.last_update.is_some()
            && u64::try_from(self.buckets_usage.len()).ok() == Some(self.buckets_count)
    }
@@ -1971,7 +1976,8 @@ impl DataUsageInfo {
    /// admin display. Partial data is accepted only with unique set states,
    /// a plan digest for every state, and at least one usable generation.
    pub fn is_valid_partial_snapshot(&self) -> bool {
-        if !self.usage_snapshot_partial
+        if self.usage_snapshot_bootstrap_pending
+            || !self.usage_snapshot_partial
            || self.usage_snapshot_converged != Some(false)
            || self.last_update.is_none()
            || self.scanner_cycle.is_none()
@@ -100,7 +100,8 @@ aws-sdk-s3 = { workspace = true, default-features = false, features = ["sigv4a",
 aws-sdk-sts = { workspace = true, default-features = false, features = ["default-https-client", "rt-tokio"] }
 aws-config = { workspace = true }
 aws-smithy-http-client = { workspace = true, default-features = false, features = ["rustls-aws-lc"] }
-async-compression = { workspace = true, features = ["tokio", "bzip2", "xz"] }
+aws-smithy-types.workspace = true
+async-compression = { workspace = true, features = ["tokio", "bzip2", "lz4", "xz"] }
 async-trait = { workspace = true }
 flate2.workspace = true
 http.workspace = true
@@ -109,15 +110,23 @@ hyper = { workspace = true, features = ["http2", "http1", "server"] }
 hyper-util = { workspace = true, features = ["tokio", "server-auto", "server-graceful", "tracing"] }
 reqwest = { workspace = true, features = ["json", "multipart", "stream"] }
 rustfs-signer.workspace = true
+
+# The MFA e2e test computes RFC 6238 codes itself rather than calling the
+# server's implementation: a shared helper could agree with a bug on both sides.
+data-encoding = { workspace = true }
+hmac = { workspace = true }
+minlz.workspace = true
+sha1 = { workspace = true }
+serde_urlencoded = { workspace = true }
 tracing = { workspace = true }
 tracing-subscriber = { workspace = true, features = ["env-filter", "time"] }
 uuid = { workspace = true, features = ["v4", "fast-rng", "macro-diagnostics"] }
 urlencoding.workspace = true
 walkdir.workspace = true
-base64 = { workspace = true }
+base64-simd = { workspace = true }
 rand = { workspace = true, features = ["serde"] }
 chrono = { workspace = true, features = ["serde"] }
-hex = { workspace = true }
+hex-simd = { workspace = true }
 md-5 = { workspace = true }
 opentelemetry-proto = { workspace = true }
 prost.workspace = true
@@ -27,6 +27,7 @@ Registered in [`src/lib.rs`](src/lib.rs). Grouped by concern:
 | **reliant** | [`src/reliant/`](src/reliant) | Tests that reuse an **externally started** server (SQL/select, conditional writes, lifecycle, deleted-object reads, node-interact). Run via [`scripts/run_e2e_tests.sh`](../../scripts/run_e2e_tests.sh); see [`src/reliant/README.md`](src/reliant/README.md) |
 | **cluster** | `cluster_concurrency_test`, `stale_multipart_cleanup_cluster_test`, `namespace_lock_quorum_test`, `admin_timeout_regression_test`, `object_lambda_test`, `replication_extension_test` | Multi-node scenarios via `RustFSTestClusterEnvironment` |
 | **chaos / reliability** | [`src/chaos.rs`](src/chaos.rs), `reliability_disk_fault_test`, `heal_erasure_disk_rebuild_test`, `server_startup_failfast_test` | Disk offline/replace/corrupt, EC rebuild, heal, fail-fast startup |
+| **upgrade compatibility** | `upgrade_compatibility_test` | Pinned previous-release writes followed by current-build reads on the same data directory |

 ## How to run

@@ -72,7 +73,7 @@ The reason string on each attribute is the classifier. Current classes:

 - **Needs a pre-started server** — `"requires running RustFS server at
  localhost:9000"` / `"Connects to existing rustfs server"`. These are the
-  `reliant/*` and `policy/test_runner` tests; start a server first (e.g.
+  `reliant/*` tests; start a server first (e.g.
  [`scripts/run_e2e_tests.sh`](../../scripts/run_e2e_tests.sh)) or use
  `--run-ignored`.
 - **Heavy / external tool** — `"Starts a rustfs server; enable when running
@@ -168,6 +169,7 @@ the same profile for membership and execution with one nightly worker.
 | `s3s-e2e` black-box | `e2e-tests` + `e2e-tests-rio-v2` jobs | **Active** (external conformance tool) |
 | ILM / lifecycle (ignored) | `test-ilm-integration-serial` lane, `-j1` | **Active** (backlog#1148 ilm-1) |
 | KMS suite | `e2e-full` job, merge queue + main | **Active** |
+| Direct and mixed-version rolling upgrades from pinned previous release | `e2e-upgrade.yml`, storage-sensitive PRs + release tags + weekly | **Active** |
 | Cluster faults (`e2e-nightly` profile) | consolidated nightly workflow | **Active** (backlog#1149 ci-7) |
 | Protocols (FTPS/WebDAV/SFTP) | consolidated nightly workflow, serial | **Active** (backlog#1149 ci-7) |
 | Replication (fast subset) | `e2e-smoke` profile, `e2e-tests` job, every PR | **Active** (backlog#1147 repl-1) |
@@ -231,8 +233,8 @@ spawn error. Install the pinned CI version before running their profiles.
  [`src/policy/README.md`](src/policy/README.md),
  [`src/protocols/README.md`](src/protocols/README.md),
  [`src/reliant/README.md`](src/reliant/README.md)
- Authoritative per-module counts:
-  [`docs/testing/e2e-suite-inventory.md`](../../docs/testing/e2e-suite-inventory.md)
+- Per-module counts: `cargo nextest list -p e2e_test --profile <profile>`
+  (one-liner in [`docs/testing/README.md`](../../docs/testing/README.md))
 - Test pyramid & flake policy: [`docs/testing/README.md`](../../docs/testing/README.md)

 ## CI smoke subset (`--profile e2e-smoke`)
@@ -269,12 +271,12 @@ Note on `#[serial]`: nextest runs each test in its own process, so
 parallel-safe by construction (random port + isolated temp dir), which the
 current subset is.

-### Authoritative test inventory
+### Test inventory

-`docs/testing/e2e-suite-inventory.md` records the per-module test counts as
-listed by `cargo nextest list -p e2e_test`. Regenerate it when adding or
-moving e2e tests so acceptance numbers in the test-strategy issues
-(backlog#1147–#1155) stay auditable. When a profile membership change is
+Per-module counts are not committed; list them with
+`cargo nextest list -p e2e_test --profile <profile>` (the result is
+platform-dependent because some modules are linux-only; the `jq` one-liner is
+in `docs/testing/README.md`). When a profile membership change is
 intentional, review its JSON listing before updating the matching
 `.config/e2e-*-selection.txt` test-ID digest. Update only the platform that
 produced the listing:
@@ -31,13 +31,9 @@

 #[cfg(test)]
 mod tests {
-    use crate::common::{RustFSTestEnvironment, init_logging, local_http_client, rustfs_binary_path};
-    use aws_sdk_s3::config::{Credentials, Region};
-    use aws_sdk_s3::{Client, Config};
-    use http::header::HOST;
-    use rustfs_signer::constants::UNSIGNED_PAYLOAD;
-    use rustfs_signer::sign_v4;
-    use s3s::Body;
+    use crate::common::{RustFSTestEnvironment, init_logging, rustfs_binary_path};
+    use aws_sdk_s3::Client;
+    use aws_sdk_s3::error::ProvideErrorMetadata;
    use std::error::Error;
    use std::io::Read;
    use std::process::{Command, Stdio};
@@ -86,10 +82,10 @@ mod tests {
    }

    /// Send a SigV4-signed request to `path` (optionally with a JSON `body`) and
-    /// return `(status, body)`. Uses the `UNSIGNED_PAYLOAD` content hash so a
-    /// request body can be attached without the caller pre-hashing it — the
-    /// server verifies the signature against the same sentinel, exactly as the
-    /// AWS SDKs / MinIO client do for streaming/unsigned payloads.
+    /// return `(status, body)`.
+    ///
+    /// Thin wrapper over [`crate::common::admin_request`], kept local so the
+    /// call sites below keep their `Option<&str>` body shape.
    async fn signed_request(
        base_url: &str,
        method: http::Method,
@@ -98,47 +94,13 @@ mod tests {
        access_key: &str,
        secret_key: &str,
    ) -> Result<(reqwest::StatusCode, String), Box<dyn Error + Send + Sync>> {
-        let url = format!("{base_url}{path}");
-        let uri = url.parse::<http::Uri>()?;
-        let authority = uri.authority().ok_or("missing authority")?.to_string();
-        let body_bytes = body.map(|b| b.as_bytes().to_vec()).unwrap_or_default();
-
-        // The signature is computed over `UNSIGNED_PAYLOAD`, so the body bytes do
-        // not participate in the SigV4 hash — sign over an empty body and attach
-        // the real payload to the wire request below.
-        let request = http::Request::builder()
-            .method(method.clone())
-            .uri(uri)
-            .header(HOST, authority)
-            .header("x-amz-content-sha256", UNSIGNED_PAYLOAD);
-        let signed = sign_v4(request.body(Body::empty())?, 0, access_key, secret_key, "", "us-east-1");
-
-        let client = local_http_client();
-        let mut rb = client.request(method, url.as_str());
-        for (name, value) in signed.headers() {
-            rb = rb.header(name, value);
-        }
-        if !body_bytes.is_empty() {
-            rb = rb.body(body_bytes);
-        }
-        let resp = rb.send().await?;
-        let status = resp.status();
-        let text = resp.text().await?;
-        Ok((status, text))
+        crate::common::admin_request(base_url, method, path, body.map(str::to_string), access_key, secret_key).await
    }

    /// Build an S3 client bound to explicit credentials (used to exercise the S3
    /// data plane with rotated / stale root credentials).
    fn s3_client_with(env: &RustFSTestEnvironment, access_key: &str, secret_key: &str) -> Client {
-        let credentials = Credentials::new(access_key, secret_key, None, None, "sec4-admin-auth");
-        let config = Config::builder()
-            .credentials_provider(credentials)
-            .region(Region::new("us-east-1"))
-            .endpoint_url(&env.url)
-            .force_path_style(true)
-            .behavior_version_latest()
-            .build();
-        Client::from_conf(config)
+        env.create_s3_client_with_credentials(access_key, secret_key)
    }

    /// Create a non-admin IAM user via the admin `add-user` API using the root
@@ -150,12 +112,7 @@ mod tests {
        access_key: &str,
        secret_key: &str,
    ) -> Result<(), Box<dyn Error + Send + Sync>> {
-        let path = format!("/rustfs/admin/v3/add-user?accessKey={access_key}");
-        let body = serde_json::json!({ "secretKey": secret_key, "status": "enabled" }).to_string();
-        let (status, resp) =
-            signed_request(&env.url, http::Method::PUT, &path, Some(&body), &env.access_key, &env.secret_key).await?;
-        assert!(status.is_success(), "add-user should succeed (status={status}, body={resp})");
-        Ok(())
+        crate::common::admin_create_user(env, access_key, secret_key).await
    }

    /// A fully authenticated but non-admin credential must be rejected with
@@ -368,10 +325,15 @@ mod tests {
            reqwest::StatusCode::FORBIDDEN,
            "stale root must be rejected on the admin API after rotation, body: {body}"
        );
-        let s3_old = s3_client_with(&env, &old_ak, &old_sk).list_buckets().send().await;
-        assert!(
-            s3_old.is_err(),
-            "stale root must be rejected on the S3 plane after rotation, got: {s3_old:?}"
+        let s3_old = s3_client_with(&env, &old_ak, &old_sk)
+            .list_buckets()
+            .send()
+            .await
+            .expect_err("stale root must be rejected on the S3 plane after rotation");
+        assert_eq!(
+            s3_old.as_service_error().and_then(ProvideErrorMetadata::code),
+            Some("InvalidAccessKeyId"),
+            "stale root must receive InvalidAccessKeyId after rotation: {s3_old:?}"
        );

        env.stop_server();
@@ -30,6 +30,7 @@ use crate::common::{
    RustFSTestEnvironment, admin_ok, admin_request, admin_request_with_session_token, build_test_sts_client, init_logging,
 };
 use aws_sdk_s3::config::{Credentials, Region};
+use aws_sdk_s3::error::ProvideErrorMetadata;
 use aws_sdk_s3::primitives::ByteStream;
 use aws_sdk_s3::{Client, Config};
 use reqwest::StatusCode;
@@ -411,8 +412,13 @@ async fn test_admin_user_policy_service_account_crud_lifecycle() -> TestResult {
        .key("before-attach")
        .body(ByteStream::from_static(b"x"))
        .send()
-        .await;
-    assert!(denied.is_err(), "user without a policy must not be able to write to {bucket}");
+        .await
+        .expect_err("user without a policy must not be able to write to the bucket");
+    assert_eq!(
+        denied.as_service_error().and_then(ProvideErrorMetadata::code),
+        Some("AccessDenied"),
+        "user without a policy must receive AccessDenied: {denied:?}"
+    );

    // --- attach policy: the credential actually gains S3 access -----------------
    admin_ok(
@@ -499,13 +505,19 @@ async fn test_admin_user_policy_service_account_crud_lifecycle() -> TestResult {
            .body(ByteStream::from_static(b"x"))
            .send()
            .await;
-        if revoked.is_err() {
-            break;
+        match revoked {
+            Ok(_) if tokio::time::Instant::now() >= deadline => {
+                return Err("deleted service account credential still works".into());
+            }
+            Ok(_) => sleep(Duration::from_millis(500)).await,
+            Err(error) => {
+                let code = error.as_service_error().and_then(ProvideErrorMetadata::code);
+                if matches!(code, Some("AccessDenied" | "InvalidAccessKeyId")) {
+                    break;
+                }
+                return Err(format!("deleted service account must fail with an authorization error, got {error:?}").into());
+            }
        }
-        if tokio::time::Instant::now() >= deadline {
-            return Err("deleted service account credential still works".into());
-        }
-        sleep(Duration::from_millis(500)).await;
    }

    // Disable then remove the user; the credential must stop working.
@@ -525,13 +537,19 @@ async fn test_admin_user_policy_service_account_crud_lifecycle() -> TestResult {
            .body(ByteStream::from_static(b"x"))
            .send()
            .await;
-        if disabled.is_err() {
-            break;
+        match disabled {
+            Ok(_) if tokio::time::Instant::now() >= deadline => {
+                return Err("disabled user credential still works".into());
+            }
+            Ok(_) => sleep(Duration::from_millis(500)).await,
+            Err(error) => {
+                let code = error.as_service_error().and_then(ProvideErrorMetadata::code);
+                if matches!(code, Some("AccessDenied" | "InvalidAccessKeyId")) {
+                    break;
+                }
+                return Err(format!("disabled user must fail with an authorization error, got {error:?}").into());
+            }
        }
-        if tokio::time::Instant::now() >= deadline {
-            return Err("disabled user credential still works".into());
-        }
-        sleep(Duration::from_millis(500)).await;
    }

    admin_ok(
@@ -0,0 +1,426 @@
+// Copyright 2024 RustFS Team
+//
+// Licensed under the Apache License, Version 2.0 (the "License");
+// you may not use this file except in compliance with the License.
+// You may obtain a copy of the License at
+//
+//     http://www.apache.org/licenses/LICENSE-2.0
+//
+// Unless required by applicable law or agreed to in writing, software
+// distributed under the License is distributed on an "AS IS" BASIS,
+// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+// See the License for the specific language governing permissions and
+// limitations under the License.
+
+//! End-to-end coverage for the self-service account and two-factor surface.
+//!
+//! The unit tests cover the state machine at its edges; what only an end-to-end
+//! test can prove is that the pieces are wired together and that the *existing*
+//! authentication paths still behave. Specifically:
+//!
+//! 1. Enrollment is refused when `RUSTFS_IAM_MASTER_KEY` is absent, so a TOTP
+//!    secret is never written where an attacker could read it off a disk.
+//! 2. With a master key, the full flow works: enroll, activate with a real
+//!    RFC 6238 code, and receive single-use recovery codes.
+//! 3. Once a factor is enrolled, `AssumeRole` refuses to mint a session without
+//!    one, and accepts a valid code — the actual login gate.
+//! 4. A direct SigV4 admin request keeps working with a factor enrolled. This is
+//!    the regression that matters most: gating it would break every script and
+//!    CLI the moment somebody enabled 2FA.
+//! 5. `AssumeRole` for an identity with no enrollment is byte-for-byte the old
+//!    behaviour, so existing deployments are untouched.
+//! 6. Rotating a password through `/account/password` invalidates the sessions
+//!    minted under the old secret.
+
+#[cfg(test)]
+mod tests {
+    use crate::common::{RustFSTestEnvironment, init_logging, local_http_client};
+    use hmac::{Hmac, KeyInit as _, Mac};
+    use http::header::HOST;
+    use rustfs_signer::constants::UNSIGNED_PAYLOAD;
+    use rustfs_signer::sign_v4;
+    use s3s::Body;
+    use sha1::Sha1;
+    use std::error::Error;
+    use std::time::{SystemTime, UNIX_EPOCH};
+
+    const ACCOUNT_INFO_PATH: &str = "/rustfs/admin/v3/account/info";
+    const ACCOUNT_PASSWORD_PATH: &str = "/rustfs/admin/v3/account/password";
+    const ACCOUNT_MFA_PATH: &str = "/rustfs/admin/v3/account/mfa";
+    const ACCOUNT_MFA_ENROLL_PATH: &str = "/rustfs/admin/v3/account/mfa/enroll";
+    const ACCOUNT_MFA_ACTIVATE_PATH: &str = "/rustfs/admin/v3/account/mfa/activate";
+    const MFA_CHALLENGE_PATH: &str = "/rustfs/admin/v3/mfa/challenge";
+    const ADMIN_INFO_PATH: &str = "/rustfs/admin/v3/info";
+
+    /// A master key so the server will accept an enrollment. Test-only value.
+    const TEST_MASTER_KEY: &str = "e2e-mfa-master-key-do-not-reuse";
+
+    type HmacSha1 = Hmac<Sha1>;
+
+    /// One signed admin request, returning the status and the raw body.
+    ///
+    /// Thin wrapper over [`crate::common::admin_request`], kept local so the
+    /// call sites below keep their `Option<&str>` body shape.
+    async fn signed_request(
+        base_url: &str,
+        method: http::Method,
+        path: &str,
+        body: Option<&str>,
+        access_key: &str,
+        secret_key: &str,
+    ) -> Result<(reqwest::StatusCode, String), Box<dyn Error + Send + Sync>> {
+        crate::common::admin_request(base_url, method, path, body.map(str::to_string), access_key, secret_key).await
+    }
+
+    /// A SigV4-signed `AssumeRole` form POST, optionally carrying a second factor.
+    ///
+    /// Uses STS's own `SerialNumber`/`TokenCode` fields, which is the point: a
+    /// script or SDK can present the factor without a RustFS-specific protocol.
+    async fn assume_role(
+        base_url: &str,
+        access_key: &str,
+        secret_key: &str,
+        second_factor: Option<(&str, &str)>,
+    ) -> Result<(reqwest::StatusCode, String), Box<dyn Error + Send + Sync>> {
+        let mut form = vec![
+            ("Action", "AssumeRole".to_string()),
+            ("Version", "2011-06-15".to_string()),
+            ("RoleArn", "arn:aws:iam::*:role/Admin".to_string()),
+            ("RoleSessionName", "e2e".to_string()),
+            ("DurationSeconds", "3600".to_string()),
+        ];
+        if let Some((challenge, code)) = second_factor {
+            form.push(("SerialNumber", challenge.to_string()));
+            form.push(("TokenCode", code.to_string()));
+        }
+        let body = serde_urlencoded::to_string(&form)?;
+
+        let uri = base_url.parse::<http::Uri>()?;
+        let authority = uri.authority().ok_or("missing authority")?.to_string();
+        let request = http::Request::builder()
+            .method(http::Method::POST)
+            .uri(format!("{base_url}/"))
+            .header(HOST, authority)
+            .header("content-type", "application/x-www-form-urlencoded")
+            .header("x-amz-content-sha256", UNSIGNED_PAYLOAD);
+        let signed = sign_v4(request.body(Body::empty())?, 0, access_key, secret_key, "", "us-east-1");
+
+        let client = local_http_client();
+        let mut builder = client.request(http::Method::POST, format!("{base_url}/"));
+        for (name, value) in signed.headers() {
+            builder = builder.header(name, value);
+        }
+        let response = builder.body(body).send().await?;
+        let status = response.status();
+        let text = response.text().await?;
+        Ok((status, text))
+    }
+
+    /// Generate the current RFC 6238 code for a base32 secret.
+    ///
+    /// Computed independently of the server implementation: a shared helper
+    /// could agree with a bug on both sides.
+    fn totp_now(secret_base32: &str) -> String {
+        let secret = data_encoding::BASE32_NOPAD
+            .decode(secret_base32.as_bytes())
+            .expect("server must return unpadded base32");
+        let step = SystemTime::now()
+            .duration_since(UNIX_EPOCH)
+            .expect("clock after the epoch")
+            .as_secs()
+            / 30;
+
+        let mut mac = HmacSha1::new_from_slice(&secret).expect("HMAC accepts any key length");
+        mac.update(&step.to_be_bytes());
+        let digest = mac.finalize().into_bytes();
+
+        let offset = (digest[digest.len() - 1] & 0x0f) as usize;
+        let binary = u32::from_be_bytes([
+            digest[offset] & 0x7f,
+            digest[offset + 1],
+            digest[offset + 2],
+            digest[offset + 3],
+        ]);
+        format!("{:06}", binary % 1_000_000)
+    }
+
+    fn json(body: &str) -> serde_json::Value {
+        serde_json::from_str(body).unwrap_or_else(|error| panic!("expected JSON, got {body}: {error}"))
+    }
+
+    #[tokio::test]
+    async fn enrollment_is_refused_without_at_rest_protection() -> Result<(), Box<dyn Error + Send + Sync>> {
+        init_logging();
+        let mut env = RustFSTestEnvironment::new().await?;
+        // Deliberately no RUSTFS_IAM_MASTER_KEY.
+        env.start_rustfs_server(vec![]).await?;
+
+        let (access_key, secret_key) = (env.access_key.clone(), env.secret_key.clone());
+
+        // The account surface itself works.
+        let (status, body) =
+            signed_request(&env.url, http::Method::GET, ACCOUNT_INFO_PATH, None, &access_key, &secret_key).await?;
+        assert_eq!(status, reqwest::StatusCode::OK, "account info must be reachable, body: {body}");
+        let info = json(&body);
+        assert_eq!(info["access_key"], access_key.as_str());
+        assert_eq!(info["identity_type"], "root");
+        assert_eq!(info["credentials_source"], "env");
+        // Root credentials come from a process-wide OnceLock that also derives
+        // the internode RPC secret, so they are immutable at runtime.
+        assert_eq!(info["mutable"]["password"], false);
+
+        // Status reports the refusal rather than pretending enrollment is possible.
+        let (status, body) =
+            signed_request(&env.url, http::Method::GET, ACCOUNT_MFA_PATH, None, &access_key, &secret_key).await?;
+        assert_eq!(status, reqwest::StatusCode::OK, "mfa status must be reachable, body: {body}");
+        let mfa = json(&body);
+        assert_eq!(mfa["enabled"], false);
+        assert_eq!(mfa["enrollment_available"], false);
+        assert!(
+            mfa["enrollment_blocked_reason"]
+                .as_str()
+                .is_some_and(|reason| reason.contains("RUSTFS_IAM_MASTER_KEY")),
+            "the refusal must name the variable an operator has to set, body: {body}"
+        );
+
+        // And enrolling actually fails, rather than writing a plaintext secret.
+        let (status, body) = signed_request(
+            &env.url,
+            http::Method::POST,
+            ACCOUNT_MFA_ENROLL_PATH,
+            Some("{}"),
+            &access_key,
+            &secret_key,
+        )
+        .await?;
+        assert!(
+            status.is_client_error() || status.is_server_error(),
+            "enrollment must fail without a master key, status: {status}, body: {body}"
+        );
+        assert!(
+            body.contains("RUSTFS_IAM_MASTER_KEY"),
+            "the failure must explain the remedy, body: {body}"
+        );
+
+        env.stop_server();
+        Ok(())
+    }
+
+    #[tokio::test]
+    async fn assume_role_is_unchanged_for_an_identity_with_no_second_factor() -> Result<(), Box<dyn Error + Send + Sync>> {
+        // The regression that protects every existing deployment: an identity
+        // with no enrollment must take no new code path.
+        init_logging();
+        let mut env = RustFSTestEnvironment::new().await?;
+        env.start_rustfs_server_with_env(vec![], &[("RUSTFS_IAM_MASTER_KEY", TEST_MASTER_KEY)])
+            .await?;
+
+        let (access_key, secret_key) = (env.access_key.clone(), env.secret_key.clone());
+
+        let (status, body) =
+            signed_request(&env.url, http::Method::GET, MFA_CHALLENGE_PATH, None, &access_key, &secret_key).await?;
+        assert_eq!(status, reqwest::StatusCode::OK, "challenge must be reachable, body: {body}");
+        let challenge = json(&body);
+        assert_eq!(challenge["required"], false, "no enrollment means no challenge");
+        assert!(challenge["challenge"].is_null());
+
+        let (status, body) = assume_role(&env.url, &access_key, &secret_key, None).await?;
+        assert_eq!(status, reqwest::StatusCode::OK, "AssumeRole must still work, body: {body}");
+        assert!(body.contains("<AccessKeyId>"), "expected STS credentials, body: {body}");
+
+        env.stop_server();
+        Ok(())
+    }
+
+    #[tokio::test]
+    async fn the_full_second_factor_lifecycle_gates_only_session_minting() -> Result<(), Box<dyn Error + Send + Sync>> {
+        init_logging();
+        let mut env = RustFSTestEnvironment::new().await?;
+        env.start_rustfs_server_with_env(vec![], &[("RUSTFS_IAM_MASTER_KEY", TEST_MASTER_KEY)])
+            .await?;
+
+        let (access_key, secret_key) = (env.access_key.clone(), env.secret_key.clone());
+
+        // --- Enroll ---
+        let (status, body) = signed_request(
+            &env.url,
+            http::Method::POST,
+            ACCOUNT_MFA_ENROLL_PATH,
+            Some("{}"),
+            &access_key,
+            &secret_key,
+        )
+        .await?;
+        assert_eq!(status, reqwest::StatusCode::OK, "enrollment must succeed, body: {body}");
+        let enrollment = json(&body);
+        let secret_base32 = enrollment["secret_base32"].as_str().expect("secret").to_string();
+        assert!(
+            enrollment["otpauth_uri"]
+                .as_str()
+                .is_some_and(|uri| uri.starts_with("otpauth://totp/RustFS:")),
+            "body: {body}"
+        );
+        assert!(!enrollment["qr_svg"].as_str().unwrap_or_default().is_empty(), "expected an SVG");
+        assert!(!enrollment["qr_utf8"].as_str().unwrap_or_default().is_empty(), "expected block art");
+
+        // A pending enrollment must not gate anything yet: a mis-scanned QR
+        // cannot be allowed to lock the operator out.
+        let (status, body) =
+            signed_request(&env.url, http::Method::GET, MFA_CHALLENGE_PATH, None, &access_key, &secret_key).await?;
+        assert_eq!(status, reqwest::StatusCode::OK);
+        assert_eq!(json(&body)["required"], false, "a pending enrollment must not gate login");
+
+        // --- Activate ---
+        let code = totp_now(&secret_base32);
+        let (status, body) = signed_request(
+            &env.url,
+            http::Method::POST,
+            ACCOUNT_MFA_ACTIVATE_PATH,
+            Some(&format!(r#"{{"code":"{code}"}}"#)),
+            &access_key,
+            &secret_key,
+        )
+        .await?;
+        assert_eq!(status, reqwest::StatusCode::OK, "activation must succeed, body: {body}");
+        let activated = json(&body);
+        let recovery_codes = activated["recovery_codes"].as_array().expect("recovery codes").clone();
+        assert_eq!(recovery_codes.len(), 10, "expected a full recovery set, body: {body}");
+
+        // --- The gate is now on for session minting ---
+        let (status, body) =
+            signed_request(&env.url, http::Method::GET, MFA_CHALLENGE_PATH, None, &access_key, &secret_key).await?;
+        assert_eq!(status, reqwest::StatusCode::OK);
+        let challenge_body = json(&body);
+        assert_eq!(challenge_body["required"], true, "body: {body}");
+        let challenge = challenge_body["challenge"].as_str().expect("challenge").to_string();
+
+        let (status, body) = assume_role(&env.url, &access_key, &secret_key, None).await?;
+        assert!(status.is_client_error(), "AssumeRole must refuse without a factor, body: {body}");
+        assert!(
+            body.contains("MultiFactorAuthRequired"),
+            "clients match on this code to prompt instead of reporting a failed login, body: {body}"
+        );
+
+        // --- ... but direct SigV4 access is untouched ---
+        let (status, body) = signed_request(&env.url, http::Method::GET, ADMIN_INFO_PATH, None, &access_key, &secret_key).await?;
+        assert_eq!(
+            status,
+            reqwest::StatusCode::OK,
+            "a direct admin request must keep working with a factor enrolled, body: {body}"
+        );
+
+        // --- A valid factor mints the session ---
+        // A fresh code: activation consumed the previous time step, so reusing
+        // that code would be refused as a replay.
+        let code = wait_for_a_fresh_code(&secret_base32).await;
+        let (status, body) = assume_role(&env.url, &access_key, &secret_key, Some((&challenge, &code))).await?;
+        assert_eq!(status, reqwest::StatusCode::OK, "a valid factor must mint a session, body: {body}");
+        assert!(body.contains("<AccessKeyId>"), "expected STS credentials, body: {body}");
+
+        // --- A recovery code also works, once ---
+        let recovery_code = recovery_codes[0].as_str().expect("recovery code").to_string();
+        let (status, body) = assume_role(&env.url, &access_key, &secret_key, Some((&challenge, &recovery_code))).await?;
+        assert_eq!(status, reqwest::StatusCode::OK, "a recovery code must mint a session, body: {body}");
+
+        let (status, body) = assume_role(&env.url, &access_key, &secret_key, Some((&challenge, &recovery_code))).await?;
+        assert!(
+            status.is_client_error(),
+            "a spent recovery code must not work twice, status: {status}, body: {body}"
+        );
+
+        env.stop_server();
+        Ok(())
+    }
+
+    #[tokio::test]
+    async fn an_iam_user_can_rotate_its_own_password_and_lose_its_sessions() -> Result<(), Box<dyn Error + Send + Sync>> {
+        init_logging();
+        let mut env = RustFSTestEnvironment::new().await?;
+        env.start_rustfs_server_with_env(vec![], &[("RUSTFS_IAM_MASTER_KEY", TEST_MASTER_KEY)])
+            .await?;
+
+        let (root_ak, root_sk) = (env.access_key.clone(), env.secret_key.clone());
+        let user_ak = "mfarotationuser";
+        let old_sk = "mfarotationsecret";
+        let new_sk = "mfarotationsecret2";
+
+        // Root creates the user.
+        let (status, body) = signed_request(
+            &env.url,
+            http::Method::PUT,
+            &format!("/rustfs/admin/v3/add-user?accessKey={user_ak}"),
+            Some(&format!(r#"{{"secretKey":"{old_sk}","status":"enabled"}}"#)),
+            &root_ak,
+            &root_sk,
+        )
+        .await?;
+        assert_eq!(status, reqwest::StatusCode::OK, "user creation must succeed, body: {body}");
+
+        // The user sees itself as mutable, unlike root.
+        let (status, body) = signed_request(&env.url, http::Method::GET, ACCOUNT_INFO_PATH, None, user_ak, old_sk).await?;
+        assert_eq!(status, reqwest::StatusCode::OK, "body: {body}");
+        let info = json(&body);
+        assert_eq!(info["identity_type"], "iam");
+        assert_eq!(info["credentials_source"], "iam");
+        assert_eq!(info["mutable"]["password"], true);
+
+        // The wrong current secret is refused, so a live session alone cannot
+        // rewrite the credential.
+        let (status, body) = signed_request(
+            &env.url,
+            http::Method::POST,
+            ACCOUNT_PASSWORD_PATH,
+            Some(&format!(r#"{{"current_secret_key":"wrong-secret","new_secret_key":"{new_sk}"}}"#)),
+            user_ak,
+            old_sk,
+        )
+        .await?;
+        assert!(status.is_client_error(), "a wrong current secret must be refused, body: {body}");
+
+        // The correct one rotates it.
+        let (status, body) = signed_request(
+            &env.url,
+            http::Method::POST,
+            ACCOUNT_PASSWORD_PATH,
+            Some(&format!(r#"{{"current_secret_key":"{old_sk}","new_secret_key":"{new_sk}"}}"#)),
+            user_ak,
+            old_sk,
+        )
+        .await?;
+        assert_eq!(status, reqwest::StatusCode::OK, "rotation must succeed, body: {body}");
+
+        // The new secret works and the old one does not.
+        let (status, body) = signed_request(&env.url, http::Method::GET, ACCOUNT_INFO_PATH, None, user_ak, new_sk).await?;
+        assert_eq!(status, reqwest::StatusCode::OK, "the new secret must work, body: {body}");
+
+        let (status, _) = signed_request(&env.url, http::Method::GET, ACCOUNT_INFO_PATH, None, user_ak, old_sk).await?;
+        assert!(status.is_client_error(), "the old secret must stop working, status: {status}");
+
+        env.stop_server();
+        Ok(())
+    }
+
+    /// Wait until the current time step differs from the one a code was just
+    /// consumed in, then return a code for it.
+    ///
+    /// Anti-replay burns the step, so a test that reuses a code inside its own
+    /// window would fail for the right reason at the wrong moment.
+    async fn wait_for_a_fresh_code(secret_base32: &str) -> String {
+        let step_at_start = SystemTime::now()
+            .duration_since(UNIX_EPOCH)
+            .expect("clock after the epoch")
+            .as_secs()
+            / 30;
+
+        loop {
+            let now = SystemTime::now()
+                .duration_since(UNIX_EPOCH)
+                .expect("clock after the epoch")
+                .as_secs();
+            if now / 30 > step_at_start {
+                return totp_now(secret_base32);
+            }
+            tokio::time::sleep(std::time::Duration::from_millis(500)).await;
+        }
+    }
+}
@@ -16,8 +16,10 @@

 #[cfg(test)]
 mod tests {
+    use std::borrow::Borrow;
+
    use crate::common::{RustFSTestEnvironment, init_logging, signed_s3_request};
-    use aws_sdk_s3::error::ProvideErrorMetadata;
+    use aws_sdk_s3::error::{ProvideErrorMetadata, SdkError};
    use aws_sdk_s3::types::{
        AccelerateConfiguration, BucketAccelerateStatus, BucketLoggingStatus, IndexDocument, LoggingEnabled, Payer,
        RequestPaymentConfiguration, WebsiteConfiguration,
@@ -26,6 +28,26 @@ mod tests {
    use http::header::CONTENT_TYPE;
    use tracing::info;

+    fn assert_s3_error<T, E, R>(result: Result<T, R>, expected_status: u16, expected_code: &str, context: &str)
+    where
+        T: std::fmt::Debug,
+        E: ProvideErrorMetadata + std::fmt::Debug,
+        R: Borrow<SdkError<E>> + std::fmt::Debug,
+    {
+        let error = result.expect_err(context);
+        let sdk_error = error.borrow();
+        assert_eq!(
+            sdk_error.raw_response().map(|response| response.status().as_u16()),
+            Some(expected_status),
+            "{context}: expected HTTP {expected_status}, got: {error:?}"
+        );
+        assert_eq!(
+            sdk_error.as_service_error().and_then(ProvideErrorMetadata::code),
+            Some(expected_code),
+            "{context}: expected {expected_code}, got: {error:?}"
+        );
+    }
+
    #[tokio::test]
    async fn test_dummy_bucket_compatibility_endpoints() {
        init_logging();
@@ -217,17 +239,11 @@ mod tests {
            .expect("DeleteBucketWebsite should return success");

        let website_after_delete = client.get_bucket_website().bucket(bucket).send().await;
-        assert!(
-            website_after_delete.is_err(),
-            "GetBucketWebsite should return NoSuchWebsiteConfiguration after deletion"
-        );
-        let website_err = website_after_delete.err().unwrap();
-        let website_code = website_err.as_service_error().and_then(|e| e.code());
-        assert!(
-            matches!(website_code, Some("NoSuchWebsiteConfiguration")),
-            "Unexpected GetBucketWebsite error code: {:?}, err: {:?}",
-            website_code,
-            website_err
+        assert_s3_error(
+            website_after_delete,
+            404,
+            "NoSuchWebsiteConfiguration",
+            "GetBucketWebsite after deleting the website configuration",
        );

        env.stop_server();
@@ -245,15 +261,7 @@ mod tests {
        let missing_bucket = "test-dummy-bucket-missing";

        let get_logging = client.get_bucket_logging().bucket(missing_bucket).send().await;
-        assert!(get_logging.is_err(), "GetBucketLogging should fail for missing bucket");
-        let get_logging_err = get_logging.err().unwrap();
-        let get_logging_code = get_logging_err.as_service_error().and_then(|e| e.code());
-        assert!(
-            matches!(get_logging_code, Some("NoSuchBucket")),
-            "Unexpected GetBucketLogging error code: {:?}, err: {:?}",
-            get_logging_code,
-            get_logging_err
-        );
+        assert_s3_error(get_logging, 404, "NoSuchBucket", "GetBucketLogging for a missing bucket");

        let put_logging = client
            .put_bucket_logging()
@@ -261,41 +269,22 @@ mod tests {
            .bucket_logging_status(BucketLoggingStatus::builder().build())
            .send()
            .await;
-        assert!(put_logging.is_err(), "PutBucketLogging should fail for missing bucket");
-        let put_logging_err = put_logging.err().unwrap();
-        let put_logging_code = put_logging_err.as_service_error().and_then(|e| e.code());
-        assert!(
-            matches!(put_logging_code, Some("NoSuchBucket")),
-            "Unexpected PutBucketLogging error code: {:?}, err: {:?}",
-            put_logging_code,
-            put_logging_err
-        );
+        assert_s3_error(put_logging, 404, "NoSuchBucket", "PutBucketLogging for a missing bucket");

        let get_accelerate = client
            .get_bucket_accelerate_configuration()
            .bucket(missing_bucket)
            .send()
            .await;
-        assert!(get_accelerate.is_err(), "GetBucketAccelerateConfiguration should fail for missing bucket");
-        let get_accelerate_err = get_accelerate.err().unwrap();
-        let get_accelerate_code = get_accelerate_err.as_service_error().and_then(|e| e.code());
-        assert!(
-            matches!(get_accelerate_code, Some("NoSuchBucket")),
-            "Unexpected GetBucketAccelerateConfiguration error code: {:?}, err: {:?}",
-            get_accelerate_code,
-            get_accelerate_err
+        assert_s3_error(
+            get_accelerate,
+            404,
+            "NoSuchBucket",
+            "GetBucketAccelerateConfiguration for a missing bucket",
        );

        let get_request_payment = client.get_bucket_request_payment().bucket(missing_bucket).send().await;
-        assert!(get_request_payment.is_err(), "GetBucketRequestPayment should fail for missing bucket");
-        let get_request_payment_err = get_request_payment.err().unwrap();
-        let get_request_payment_code = get_request_payment_err.as_service_error().and_then(|e| e.code());
-        assert!(
-            matches!(get_request_payment_code, Some("NoSuchBucket")),
-            "Unexpected GetBucketRequestPayment error code: {:?}, err: {:?}",
-            get_request_payment_code,
-            get_request_payment_err
-        );
+        assert_s3_error(get_request_payment, 404, "NoSuchBucket", "GetBucketRequestPayment for a missing bucket");

        let put_accelerate = client
            .put_bucket_accelerate_configuration()
@@ -307,14 +296,11 @@ mod tests {
            )
            .send()
            .await;
-        assert!(put_accelerate.is_err(), "PutBucketAccelerateConfiguration should fail for missing bucket");
-        let put_accelerate_err = put_accelerate.err().unwrap();
-        let put_accelerate_code = put_accelerate_err.as_service_error().and_then(|e| e.code());
-        assert!(
-            matches!(put_accelerate_code, Some("NoSuchBucket")),
-            "Unexpected PutBucketAccelerateConfiguration error code: {:?}, err: {:?}",
-            put_accelerate_code,
-            put_accelerate_err
+        assert_s3_error(
+            put_accelerate,
+            404,
+            "NoSuchBucket",
+            "PutBucketAccelerateConfiguration for a missing bucket",
        );

        let put_request_payment = client
@@ -328,15 +314,7 @@ mod tests {
            )
            .send()
            .await;
-        assert!(put_request_payment.is_err(), "PutBucketRequestPayment should fail for missing bucket");
-        let put_request_payment_err = put_request_payment.err().unwrap();
-        let put_request_payment_code = put_request_payment_err.as_service_error().and_then(|e| e.code());
-        assert!(
-            matches!(put_request_payment_code, Some("NoSuchBucket")),
-            "Unexpected PutBucketRequestPayment error code: {:?}, err: {:?}",
-            put_request_payment_code,
-            put_request_payment_err
-        );
+        assert_s3_error(put_request_payment, 404, "NoSuchBucket", "PutBucketRequestPayment for a missing bucket");

        let put_website = client
            .put_bucket_website()
@@ -353,37 +331,13 @@ mod tests {
            )
            .send()
            .await;
-        assert!(put_website.is_err(), "PutBucketWebsite should fail for missing bucket");
-        let put_website_err = put_website.err().unwrap();
-        let put_website_code = put_website_err.as_service_error().and_then(|e| e.code());
-        assert!(
-            matches!(put_website_code, Some("NoSuchBucket")),
-            "Unexpected PutBucketWebsite error code: {:?}, err: {:?}",
-            put_website_code,
-            put_website_err
-        );
+        assert_s3_error(put_website, 404, "NoSuchBucket", "PutBucketWebsite for a missing bucket");

        let get_website = client.get_bucket_website().bucket(missing_bucket).send().await;
-        assert!(get_website.is_err(), "GetBucketWebsite should fail for missing bucket");
-        let get_website_err = get_website.err().unwrap();
-        let get_website_code = get_website_err.as_service_error().and_then(|e| e.code());
-        assert!(
-            matches!(get_website_code, Some("NoSuchBucket")),
-            "Unexpected GetBucketWebsite error code: {:?}, err: {:?}",
-            get_website_code,
-            get_website_err
-        );
+        assert_s3_error(get_website, 404, "NoSuchBucket", "GetBucketWebsite for a missing bucket");

        let delete_website = client.delete_bucket_website().bucket(missing_bucket).send().await;
-        assert!(delete_website.is_err(), "DeleteBucketWebsite should fail for missing bucket");
-        let delete_website_err = delete_website.err().unwrap();
-        let delete_website_code = delete_website_err.as_service_error().and_then(|e| e.code());
-        assert!(
-            matches!(delete_website_code, Some("NoSuchBucket")),
-            "Unexpected DeleteBucketWebsite error code: {:?}, err: {:?}",
-            delete_website_code,
-            delete_website_err
-        );
+        assert_s3_error(delete_website, 404, "NoSuchBucket", "DeleteBucketWebsite for a missing bucket");

        env.stop_server();
    }
@@ -15,38 +15,23 @@
 //! Regression test for Issue #1423
 //! Verifies that Bucket Policies are honored for Authenticated Users.

-use crate::common::{RustFSTestEnvironment, init_logging};
-use aws_sdk_s3::config::{Credentials, Region};
-use aws_sdk_s3::{Client, Config};
+use crate::common::{AdminTransport, RustFSTestEnvironment, admin_create_user_via, init_logging};
+use aws_sdk_s3::Client;
+use aws_sdk_s3::error::ProvideErrorMetadata;
 use tracing::info;

+/// This suite deliberately drives the admin API through the external `awscurl`
+/// binary, so user creation pins `AdminTransport::Awscurl`.
 async fn create_user(
    env: &RustFSTestEnvironment,
    username: &str,
    password: &str,
 ) -> Result<(), Box<dyn std::error::Error + Send + Sync>> {
-    let create_user_body = serde_json::json!({
-        "secretKey": password,
-        "status": "enabled"
-    })
-    .to_string();
-
-    let create_user_url = format!("{}/rustfs/admin/v3/add-user?accessKey={}", env.url, username);
-    crate::common::awscurl_put(&create_user_url, &create_user_body, &env.access_key, &env.secret_key).await?;
-    Ok(())
+    admin_create_user_via(AdminTransport::Awscurl, &env.url, &env.access_key, &env.secret_key, username, password).await
 }

 fn create_user_client(env: &RustFSTestEnvironment, access_key: &str, secret_key: &str) -> Client {
-    let credentials = Credentials::new(access_key, secret_key, None, None, "test-user");
-    let config = Config::builder()
-        .credentials_provider(credentials)
-        .region(Region::new("us-east-1"))
-        .endpoint_url(&env.url)
-        .force_path_style(true)
-        .behavior_version_latest()
-        .build();
-
-    Client::from_conf(config)
+    env.create_s3_client_with_credentials(access_key, secret_key)
 }

 #[tokio::test]
@@ -73,10 +58,14 @@ async fn test_bucket_policy_authenticated_user() -> Result<(), Box<dyn std::erro
    let user_client = create_user_client(&env, user_access, user_secret);

    // 4. Verify Access Denied initially (No Policy)
-    let result = user_client.list_objects_v2().bucket(bucket_name).send().await;
-    if result.is_ok() {
-        return Err("Should be Access Denied initially".into());
-    }
+    let denied = user_client
+        .list_objects_v2()
+        .bucket(bucket_name)
+        .send()
+        .await
+        .expect_err("a user without a bucket policy must be denied");
+    assert_eq!(denied.raw_response().map(|response| response.status().as_u16()), Some(403));
+    assert_eq!(denied.as_service_error().and_then(ProvideErrorMetadata::code), Some("AccessDenied"));

    // 5. Apply Bucket Policy Allowed User
    let policy_json = serde_json::json!({
@@ -20,10 +20,10 @@ mod tests {
    use crate::common::{RustFSTestEnvironment, init_logging};
    use aws_sdk_s3::Client;
    use aws_sdk_s3::config::{Credentials, Region, RequestChecksumCalculation};
+    use aws_sdk_s3::error::ProvideErrorMetadata;
    use aws_sdk_s3::primitives::ByteStream;
-    use aws_sdk_s3::types::{ChecksumAlgorithm, ChecksumMode, CompletedMultipartUpload, CompletedPart};
+    use aws_sdk_s3::types::{ChecksumAlgorithm, ChecksumMode, CompletedMultipartUpload, CompletedPart, ServerSideEncryption};
    use aws_smithy_http_client::Builder as SmithyHttpClientBuilder;
-    use base64::Engine;
    use md5::{Digest as Md5Digest, Md5};
    use rustfs_rio::{Checksum, ChecksumType as RioChecksumType};
    use sha2::Sha256;
@@ -73,12 +73,12 @@ mod tests {
        let mut hasher = Md5::new();
        hasher.update(body);
        let digest = hasher.finalize();
-        base64::engine::general_purpose::STANDARD.encode(digest.as_slice())
+        base64_simd::STANDARD.encode_to_string(digest.as_slice())
    }

    fn checksum_sha256_base64(body: &[u8]) -> String {
        let digest = Sha256::digest(body);
-        base64::engine::general_purpose::STANDARD.encode(digest.as_slice())
+        base64_simd::STANDARD.encode_to_string(digest.as_slice())
    }

    fn checksum_crc64nvme_base64(body: &[u8]) -> String {
@@ -186,21 +186,31 @@ mod tests {
            .send()
            .await;

-        assert!(
-            result.is_err(),
-            "PutObject with a mismatched SHA256 must be rejected, but it succeeded (issue #4341)"
+        let error = result.expect_err("PutObject with a mismatched SHA256 must be rejected (issue #4341)");
+        assert_eq!(
+            error.raw_response().map(|response| response.status().as_u16()),
+            Some(400),
+            "Mismatched SHA256 must return HTTP 400, got {error:?}"
        );
-        let err = result.err().unwrap();
-        let msg = format!("{err:?}");
-        info!("PutObject correctly rejected mismatched checksum: {msg}");
-        assert!(
-            msg.contains("BadDigest") || msg.to_lowercase().contains("digest") || msg.to_lowercase().contains("checksum"),
-            "Expected a BadDigest/checksum error, got: {msg}"
+        assert_eq!(
+            error.as_service_error().and_then(ProvideErrorMetadata::code),
+            Some("BadDigest"),
+            "Mismatched SHA256 must return BadDigest, got {error:?}"
        );

        // And the object must not have been stored.
-        let head = client.head_object().bucket(bucket).key(key).send().await;
-        assert!(head.is_err(), "Object must not exist after a rejected mismatched-checksum PutObject");
+        let error = client
+            .head_object()
+            .bucket(bucket)
+            .key(key)
+            .send()
+            .await
+            .expect_err("Object must not exist after a rejected mismatched-checksum PutObject");
+        assert_eq!(
+            error.raw_response().map(|response| response.status().as_u16()),
+            Some(404),
+            "Rejected mismatched-checksum PutObject absence probe must return HTTP 404, got {error:?}"
+        );
        info!("PASSED: PutObject rejects mismatched SHA256 and stores nothing");
    }

@@ -250,6 +260,117 @@ mod tests {
        info!("PASSED: HeadObject returns stored SHA256 digest");
    }

+    #[tokio::test]
+    async fn test_head_object_returns_sse_s3_checksum() {
+        init_logging();
+
+        let mut env = RustFSTestEnvironment::new().await.expect("Failed to create test environment");
+        env.start_rustfs_server_with_env(
+            vec![],
+            &[
+                ("RUSTFS_SSE_S3_MASTER_KEY", "MTIzNDU2Nzg5MDEyMzQ1Njc4OTAxMjM0NTY3ODkwMTI="),
+                ("RUSTFS_CONSOLE_ENABLE", "false"),
+            ],
+        )
+        .await
+        .expect("Failed to start RustFS");
+
+        let client = create_s3_client(&env);
+        let bucket = "test-sse-s3-checksum-head";
+        create_bucket(&client, bucket).await.expect("Failed to create bucket");
+
+        let put = client
+            .put_object()
+            .bucket(bucket)
+            .key("encrypted.txt")
+            .body(ByteStream::from_static(b"encrypted checksum"))
+            .server_side_encryption(ServerSideEncryption::Aes256)
+            .checksum_algorithm(ChecksumAlgorithm::Crc32)
+            .send()
+            .await
+            .expect("SSE-S3 PutObject with CRC32 failed");
+        let expected = put.checksum_crc32().expect("PutObject must return CRC32");
+
+        let head = client
+            .head_object()
+            .bucket(bucket)
+            .key("encrypted.txt")
+            .checksum_mode(ChecksumMode::Enabled)
+            .send()
+            .await
+            .expect("SSE-S3 HeadObject failed");
+
+        assert_eq!(head.checksum_crc32(), Some(expected));
+
+        client
+            .copy_object()
+            .bucket(bucket)
+            .key("encrypted-copy.txt")
+            .copy_source(format!("{bucket}/encrypted.txt"))
+            .server_side_encryption(ServerSideEncryption::Aes256)
+            .send()
+            .await
+            .expect("SSE-S3 CopyObject failed");
+        let copy_head = client
+            .head_object()
+            .bucket(bucket)
+            .key("encrypted-copy.txt")
+            .checksum_mode(ChecksumMode::Enabled)
+            .send()
+            .await
+            .expect("SSE-S3 copied HeadObject failed");
+
+        assert_eq!(copy_head.checksum_crc32(), Some(expected));
+
+        let multipart_key = "encrypted-multipart.txt";
+        let create = client
+            .create_multipart_upload()
+            .bucket(bucket)
+            .key(multipart_key)
+            .server_side_encryption(ServerSideEncryption::Aes256)
+            .checksum_algorithm(ChecksumAlgorithm::Crc32)
+            .send()
+            .await
+            .expect("SSE-S3 CreateMultipartUpload with CRC32 failed");
+        let upload_id = create.upload_id().expect("CreateMultipartUpload must return an upload ID");
+        let part = client
+            .upload_part()
+            .bucket(bucket)
+            .key(multipart_key)
+            .upload_id(upload_id)
+            .part_number(1)
+            .body(ByteStream::from_static(b"encrypted multipart checksum"))
+            .checksum_algorithm(ChecksumAlgorithm::Crc32)
+            .send()
+            .await
+            .expect("SSE-S3 UploadPart with CRC32 failed");
+        let completed_part = CompletedPart::builder()
+            .part_number(1)
+            .e_tag(part.e_tag().expect("UploadPart must return an ETag"))
+            .checksum_crc32(part.checksum_crc32().expect("UploadPart must return CRC32"))
+            .build();
+        let complete = client
+            .complete_multipart_upload()
+            .bucket(bucket)
+            .key(multipart_key)
+            .upload_id(upload_id)
+            .multipart_upload(CompletedMultipartUpload::builder().parts(completed_part).build())
+            .send()
+            .await
+            .expect("SSE-S3 CompleteMultipartUpload with CRC32 failed");
+        let expected_multipart = complete.checksum_crc32().expect("CompleteMultipartUpload must return CRC32");
+        let multipart_head = client
+            .head_object()
+            .bucket(bucket)
+            .key(multipart_key)
+            .checksum_mode(ChecksumMode::Enabled)
+            .send()
+            .await
+            .expect("SSE-S3 multipart HeadObject failed");
+
+        assert_eq!(multipart_head.checksum_crc32(), Some(expected_multipart));
+    }
+
    /// Multipart upload with checksum: CreateMultipartUpload, UploadPart(s) with checksum_sha256, CompleteMultipartUpload; then GetObject verifies content.
    /// Uses part size >= 5MB (server minimum) for two parts.
    #[tokio::test]
@@ -546,14 +667,29 @@ mod tests {
                })
                .send()
                .await;
-            assert!(put_bad.is_err(), "{header}: a mismatched checksum must be rejected");
-            let msg = format!("{:?}", put_bad.err().unwrap());
-            assert!(
-                msg.contains("BadDigest") || msg.to_lowercase().contains("digest") || msg.to_lowercase().contains("checksum"),
-                "{header}: expected a BadDigest/checksum error, got: {msg}"
+            let error = put_bad.expect_err("a mismatched checksum must be rejected");
+            assert_eq!(
+                error.raw_response().map(|response| response.status().as_u16()),
+                Some(400),
+                "{header}: mismatched checksum must return HTTP 400, got {error:?}"
+            );
+            assert_eq!(
+                error.as_service_error().and_then(ProvideErrorMetadata::code),
+                Some("BadDigest"),
+                "{header}: mismatched checksum must return BadDigest, got {error:?}"
+            );
+            let error = client
+                .head_object()
+                .bucket(bucket)
+                .key(&bad_key)
+                .send()
+                .await
+                .expect_err("nothing must be stored after a rejected PutObject");
+            assert_eq!(
+                error.raw_response().map(|response| response.status().as_u16()),
+                Some(404),
+                "{header}: rejected PutObject absence probe must return HTTP 404, got {error:?}"
            );
-            let head = client.head_object().bucket(bucket).key(&bad_key).send().await;
-            assert!(head.is_err(), "{header}: nothing must be stored after a rejected PutObject");

            info!("PASSED additional-checksum verify-on-write: {header}");
        }
@@ -27,8 +27,10 @@
 //! Readiness is established by the harness's `start()` handshake (TCP reachability
 //! plus an S3 `ListBuckets` poll) — there are no fixed sleeps.
 //!
-//! Out of scope for this block (tracked separately): network fault injection
-//! (toxiproxy / socket proxy) and 5GiB large-object budgets.
+//! The volume-proxy smoke below also proves that the socket-level fault proxy
+//! can be installed before startup without changing the client-facing node URL.
+//! A full lock-plane partition matrix and 5GiB large-object budget remain
+//! tracked separately.

 use crate::common::{ClusterTopology, RustFSTestClusterEnvironment};

@@ -76,6 +78,28 @@ async fn cluster_multidrive_single_pool_smoke() -> TestResult {
    Ok(())
 }

+/// 4 nodes x 4 drives, single pool: exercise the maximum local erasure layout
+/// supported by the cluster harness. This remains in the nightly lane because
+/// it starts four real server processes and sixteen data directories.
+#[tokio::test]
+async fn cluster_four_node_four_drive_single_pool_smoke() -> TestResult {
+    crate::common::init_logging();
+
+    let mut cluster = RustFSTestClusterEnvironment::with_topology(ClusterTopology::single_pool_multidrive(4, 4)).await?;
+
+    let volumes = cluster.rustfs_volumes_arg();
+    assert_eq!(volumes.split(' ').count(), 16, "expected 16 explicit endpoints, got: {volumes}");
+    assert!(!volumes.contains('{'), "single-pool layout must not use ellipses: {volumes}");
+    assert!(cluster.nodes.iter().all(|node| node.data_dirs.len() == 4));
+
+    cluster.start().await?;
+    cluster.create_test_bucket(BUCKET).await?;
+
+    let payload = vec![0x3Cu8; 1024 * 1024];
+    put_get_roundtrip(&cluster, "multidrive-4/object", &payload).await?;
+    Ok(())
+}
+
 /// Two single-node pools, 2 drives each: the multi-pool layout boots and
 /// round-trips. Every pool is a distinct erasure pool (`pool_idx` 0 and 1).
 #[tokio::test]
@@ -103,3 +127,27 @@ async fn cluster_two_pool_smoke() -> TestResult {
    put_get_roundtrip(&cluster, "twopool/object", &payload).await?;
    Ok(())
 }
+
+/// A real cluster smoke for the volume FaultProxy wiring. The proxy target is
+/// not listening yet when it is created; cluster startup must still converge
+/// once the target node starts, and peer disk/RPC traffic must traverse it.
+#[tokio::test]
+async fn cluster_volume_fault_proxy_pass_smoke() -> TestResult {
+    crate::common::init_logging();
+
+    let mut cluster = RustFSTestClusterEnvironment::with_topology(ClusterTopology::single_pool_multidrive(2, 2)).await?;
+    let proxy = cluster.start_volume_proxy_for_node(0).await?;
+    let proxied = proxy.local_addr().to_string();
+    assert!(cluster.rustfs_volumes_arg().contains(&proxied));
+
+    let result: TestResult = async {
+        cluster.start().await?;
+        cluster.create_test_bucket(BUCKET).await?;
+        let payload = vec![0x6Du8; 256 * 1024];
+        put_get_roundtrip(&cluster, "volume-proxy/object", &payload).await
+    }
+    .await;
+
+    proxy.shutdown().await;
+    result
+}
@@ -34,6 +34,7 @@ use serde_json;
 use std::ffi::OsStr;
 use std::fs as stdfs;
 use std::io::ErrorKind;
+use std::net::SocketAddr;
 use std::path::{Path, PathBuf};
 use std::process::{Child, Command, Stdio};
 use std::sync::Once;
@@ -217,7 +218,37 @@ pub(crate) async fn signed_s3_request(
    access_key: &str,
    secret_key: &str,
 ) -> Result<reqwest::Response, Box<dyn std::error::Error + Send + Sync>> {
-    signed_s3_request_with_session_token(method, url, body, content_type, access_key, secret_key, None).await
+    signed_s3_request_with_headers(method, url, body, content_type, access_key, secret_key, &http::HeaderMap::new()).await
+}
+
+pub(crate) async fn signed_s3_request_with_headers(
+    method: http::Method,
+    url: &str,
+    body: Option<String>,
+    content_type: Option<&str>,
+    access_key: &str,
+    secret_key: &str,
+    extra_headers: &http::HeaderMap,
+) -> Result<reqwest::Response, Box<dyn std::error::Error + Send + Sync>> {
+    signed_s3_request_with_session_token(
+        method,
+        url,
+        body,
+        content_type,
+        SigningCredentials {
+            access_key,
+            secret_key,
+            session_token: None,
+        },
+        extra_headers,
+    )
+    .await
+}
+
+struct SigningCredentials<'a> {
+    access_key: &'a str,
+    secret_key: &'a str,
+    session_token: Option<&'a str>,
 }

 async fn signed_s3_request_with_session_token(
@@ -225,9 +256,8 @@ async fn signed_s3_request_with_session_token(
    url: &str,
    body: Option<String>,
    content_type: Option<&str>,
-    access_key: &str,
-    secret_key: &str,
-    session_token: Option<&str>,
+    credentials: SigningCredentials<'_>,
+    extra_headers: &http::HeaderMap,
 ) -> Result<reqwest::Response, Box<dyn std::error::Error + Send + Sync>> {
    let uri = url.parse::<http::Uri>()?;
    let authority = uri.authority().ok_or("S3 URL missing authority")?.to_string();
@@ -239,14 +269,17 @@ async fn signed_s3_request_with_session_token(
    if let Some(content_type) = content_type {
        request = request.header(CONTENT_TYPE, content_type);
    }
+    for (name, value) in extra_headers {
+        request = request.header(name, value);
+    }

    let content_length = i64::try_from(body.as_ref().map_or(0, String::len)).map_err(|_| "S3 request body is too large")?;
    let signed = sign_v4(
        request.body(Body::empty())?,
        content_length,
-        access_key,
-        secret_key,
-        session_token.unwrap_or_default(),
+        credentials.access_key,
+        credentials.secret_key,
+        credentials.session_token.unwrap_or_default(),
        "us-east-1",
    );

@@ -283,8 +316,19 @@ pub(crate) async fn admin_request_with_session_token(
 ) -> Result<(StatusCode, String), Box<dyn std::error::Error + Send + Sync>> {
    let url = format!("{base_url}{path_and_query}");
    let content_type = body.as_ref().map(|_| "application/json");
-    let response =
-        signed_s3_request_with_session_token(method, &url, body, content_type, access_key, secret_key, session_token).await?;
+    let response = signed_s3_request_with_session_token(
+        method,
+        &url,
+        body,
+        content_type,
+        SigningCredentials {
+            access_key,
+            secret_key,
+            session_token,
+        },
+        &http::HeaderMap::new(),
+    )
+    .await?;
    let status = response.status();
    let body = response.text().await?;
    Ok((status, body))
@@ -310,6 +354,17 @@ pub fn rustfs_binary_path() -> PathBuf {
    rustfs_binary_path_with_features(requested_rustfs_build_features().as_deref())
 }

+fn resolve_rustfs_binary_path(workspace: &Path, configured_target_dir: Option<&Path>) -> PathBuf {
+    let mut path = match configured_target_dir {
+        Some(path) if path.is_absolute() => path.to_path_buf(),
+        Some(path) => workspace.join(path),
+        None => workspace.join("target"),
+    };
+    path.push(if cfg!(debug_assertions) { "debug" } else { "release" });
+    path.push(format!("rustfs{}", std::env::consts::EXE_SUFFIX));
+    path
+}
+
 /// Resolve the RustFS binary relative to the workspace, optionally requesting build features.
 pub fn rustfs_binary_path_with_features(requested_features: Option<&str>) -> PathBuf {
    if let Some(path) = std::env::var_os("CARGO_BIN_EXE_rustfs") {
@@ -317,11 +372,9 @@ pub fn rustfs_binary_path_with_features(requested_features: Option<&str>) -> Pat
    }
    let requested_features = requested_features.and_then(normalize_rustfs_build_features);

-    let mut binary_path = workspace_root();
-    binary_path.push("target");
-    let profile_dir = if cfg!(debug_assertions) { "debug" } else { "release" };
-    binary_path.push(profile_dir);
-    binary_path.push(format!("rustfs{}", std::env::consts::EXE_SUFFIX));
+    let workspace = workspace_root();
+    let configured_target_dir = std::env::var_os("CARGO_TARGET_DIR").map(PathBuf::from);
+    let binary_path = resolve_rustfs_binary_path(&workspace, configured_target_dir.as_deref());

    let features_match = binary_features_match(&binary_path, requested_features.as_deref());
    let source_is_newer = workspace_sources_newer_than_binary(&binary_path);
@@ -338,7 +391,7 @@ pub fn rustfs_binary_path_with_features(requested_features: Option<&str>) -> Pat
    }

    info!("Building RustFS binary to ensure it's up to date...");
-    build_rustfs_binary(requested_features.as_deref());
+    build_rustfs_binary(requested_features.as_deref(), &binary_path);

    info!("Using RustFS binary at {:?}", binary_path);
    binary_path
@@ -440,7 +493,7 @@ fn path_is_newer_than(binary_modified: std::time::SystemTime, path: &Path) -> bo
 }

 /// Build the RustFS binary using cargo
-fn build_rustfs_binary(requested_features: Option<&str>) {
+fn build_rustfs_binary(requested_features: Option<&str>, binary_path: &Path) {
    let workspace = workspace_root();
    info!("Building RustFS binary from workspace: {:?}", workspace);

@@ -476,11 +529,7 @@ fn build_rustfs_binary(requested_features: Option<&str>) {
        panic!("Failed to build RustFS binary. Error: {stderr}");
    }

-    let mut binary_path = workspace;
-    binary_path.push("target");
-    binary_path.push(if cfg!(debug_assertions) { "debug" } else { "release" });
-    binary_path.push(format!("rustfs{}", std::env::consts::EXE_SUFFIX));
-    let stamp_path = rustfs_binary_features_stamp_path(&binary_path);
+    let stamp_path = rustfs_binary_features_stamp_path(binary_path);
    if let Err(err) = stdfs::write(&stamp_path, requested_features.unwrap_or_default()) {
        warn!("Failed to write RustFS feature stamp {:?}: {}", stamp_path, err);
    }
@@ -633,6 +682,18 @@ impl RustFSTestEnvironment {
        extra_args: Vec<&str>,
        extra_env: &[(&str, &str)],
        cleanup_existing: bool,
+    ) -> Result<(), Box<dyn std::error::Error + Send + Sync>> {
+        let binary_path = rustfs_binary_path();
+        self.start_rustfs_server_inner_with_binary(&binary_path, extra_args, extra_env, cleanup_existing)
+            .await
+    }
+
+    async fn start_rustfs_server_inner_with_binary(
+        &mut self,
+        binary_path: &Path,
+        extra_args: Vec<&str>,
+        extra_env: &[(&str, &str)],
+        cleanup_existing: bool,
    ) -> Result<(), Box<dyn std::error::Error + Send + Sync>> {
        if cleanup_existing {
            self.cleanup_existing_processes().await?;
@@ -642,8 +703,7 @@ impl RustFSTestEnvironment {

        info!("Starting RustFS server with args: {:?}", args);

-        let binary_path = rustfs_binary_path();
-        let mut command = Command::new(&binary_path);
+        let mut command = Command::new(binary_path);
        command.env("RUST_LOG", "rustfs=info,rustfs_notify=debug");
        // The embedded console would bind the fixed default port :9001, which
        // collides with unrelated local services (e.g. Docker Desktop). Tests
@@ -663,6 +723,19 @@ impl RustFSTestEnvironment {
        Ok(())
    }

+    /// Start a specific RustFS binary against this environment's isolated
+    /// data directory. Upgrade tests use this to seed an old on-disk format
+    /// before restarting the same environment with the workspace binary.
+    pub async fn start_rustfs_server_from_binary(
+        &mut self,
+        binary_path: &Path,
+        extra_args: Vec<&str>,
+        extra_env: &[(&str, &str)],
+    ) -> Result<(), Box<dyn std::error::Error + Send + Sync>> {
+        self.start_rustfs_server_inner_with_binary(binary_path, extra_args, extra_env, true)
+            .await
+    }
+
    /// Start RustFS server with basic configuration
    pub async fn start_rustfs_server(&mut self, extra_args: Vec<&str>) -> Result<(), Box<dyn std::error::Error + Send + Sync>> {
        self.start_rustfs_server_inner(extra_args, &[], true).await
@@ -1142,6 +1215,9 @@ pub struct RustFSTestClusterEnvironment {
    pub node_extra_env: Vec<Vec<(String, String)>>,
    pub node_capture_log_paths: Vec<Option<String>>,
    pub topology: ClusterTopology,
+    /// Optional socket proxies used for the corresponding node's volume
+    /// endpoints. Proxies must be installed before [`Self::start`].
+    volume_proxy_addresses: Vec<Option<SocketAddr>>,
 }

 impl RustFSTestClusterEnvironment {
@@ -1233,6 +1309,7 @@ impl RustFSTestClusterEnvironment {
            extra_env.push(("RUSTFS_UNSAFE_BYPASS_DISK_CHECK".to_string(), "true".to_string()));
        }

+        let node_count = topology.node_count;
        Ok(Self {
            nodes,
            temp_dir,
@@ -1242,6 +1319,7 @@ impl RustFSTestClusterEnvironment {
            node_extra_env: vec![Vec::new(); topology.node_count],
            node_capture_log_paths: vec![None; topology.node_count],
            topology,
+            volume_proxy_addresses: vec![None; node_count],
        })
    }

@@ -1309,6 +1387,34 @@ impl RustFSTestClusterEnvironment {
        self.build_volumes_arg()
    }

+    /// Start a socket proxy for one node's volume endpoints and route all
+    /// subsequent `RUSTFS_VOLUMES` references for that node through it.
+    ///
+    /// Call this before [`Self::start`], then use the returned proxy's
+    /// [`crate::fault_proxy::FaultProxy::set_mode`] to inject latency,
+    /// blackhole, or one-way partition faults. The node's own listen address
+    /// remains direct, so S3 clients can still reach it while peer disk/RPC
+    /// traffic is steered through the proxy.
+    pub async fn start_volume_proxy_for_node(
+        &mut self,
+        node_idx: usize,
+    ) -> Result<crate::fault_proxy::FaultProxy, Box<dyn std::error::Error + Send + Sync>> {
+        self.ensure_node_index(node_idx)?;
+        if self.volume_proxy_addresses[node_idx].is_some() {
+            return Err(format!("a volume proxy is already configured for node {node_idx}").into());
+        }
+        let target = self.nodes[node_idx].address.parse::<SocketAddr>()?;
+        let proxy = crate::fault_proxy::FaultProxy::start(target).await?;
+        self.volume_proxy_addresses[node_idx] = Some(proxy.local_addr());
+        Ok(proxy)
+    }
+
+    fn volume_address(&self, node_idx: usize) -> String {
+        self.volume_proxy_addresses[node_idx]
+            .map(|address| address.to_string())
+            .unwrap_or_else(|| self.nodes[node_idx].address.clone())
+    }
+
    fn build_volumes_arg(&self) -> String {
        let pools = self.topology.normalized_pools();

@@ -1317,7 +1423,11 @@ impl RustFSTestClusterEnvironment {
            return self
                .nodes
                .iter()
-                .flat_map(|n| n.data_dirs.iter().map(move |dir| format!("http://{}{}", n.address, dir)))
+                .enumerate()
+                .flat_map(|(node_idx, n)| {
+                    let address = self.volume_address(node_idx);
+                    n.data_dirs.iter().map(move |dir| format!("http://{}{}", address, dir))
+                })
                .collect::<Vec<_>>()
                .join(" ");
        }
@@ -1328,13 +1438,19 @@ impl RustFSTestClusterEnvironment {
        pools
            .iter()
            .map(|nodes| {
-                let node = &self.nodes[nodes[0]];
+                let node_idx = nodes[0];
+                let node = &self.nodes[node_idx];
                let base = node
                    .data_dirs
                    .first()
                    .and_then(|d| d.rsplit_once('/').map(|(parent, _)| parent))
                    .unwrap_or(&node.data_dir);
-                format!("http://{}{}/drive{{0...{}}}", node.address, base, self.topology.drives_per_node - 1)
+                format!(
+                    "http://{}{}/drive{{0...{}}}",
+                    self.volume_address(node_idx),
+                    base,
+                    self.topology.drives_per_node - 1
+                )
            })
            .collect::<Vec<_>>()
            .join(" ")
@@ -1353,31 +1469,18 @@ impl RustFSTestClusterEnvironment {
    ///   times out, or cluster service readiness times out.
    pub async fn start(&mut self) -> Result<(), Box<dyn std::error::Error + Send + Sync>> {
        let binary_path = rustfs_binary_path();
+        self.start_with_binary(&binary_path).await
+    }
+
+    /// Start every cluster node with a specific RustFS binary.
+    ///
+    /// Upgrade compatibility tests use this to initialize a cluster with a
+    /// pinned previous release before replacing nodes with the workspace build.
+    pub async fn start_with_binary(&mut self, binary_path: &Path) -> Result<(), Box<dyn std::error::Error + Send + Sync>> {
        let volumes_arg = self.build_volumes_arg();

-        for (i, node) in self.nodes.iter_mut().enumerate() {
-            info!("Starting cluster node {} on {}", i, node.address);
-
-            let mut command = Command::new(&binary_path);
-            command
-                .env("RUSTFS_VOLUMES", &volumes_arg)
-                .env("RUSTFS_ADDRESS", &node.address)
-                .env("RUSTFS_ACCESS_KEY", &self.access_key)
-                .env("RUSTFS_SECRET_KEY", &self.secret_key)
-                .env("RUSTFS_CONSOLE_ENABLE", "false")
-                .env("RUST_LOG", "rustfs=info,rustfs_notify=debug");
-
-            for (key, value) in &self.extra_env {
-                command.env(key, value);
-            }
-            for (key, value) in &self.node_extra_env[i] {
-                command.env(key, value);
-            }
-            capture_command_logs(&mut command, self.node_capture_log_paths[i].as_deref())?;
-
-            let process = command.current_dir(&node.data_dir).spawn()?;
-
-            node.process = Some(process);
+        for node_idx in 0..self.nodes.len() {
+            self.spawn_node(node_idx, binary_path, &volumes_arg)?;
        }

        for (i, node) in self.nodes.iter().enumerate() {
@@ -1393,20 +1496,46 @@ impl RustFSTestClusterEnvironment {

    /// Start one node process using the cluster's existing volume layout.
    pub async fn start_node(&mut self, node_idx: usize) -> Result<(), Box<dyn std::error::Error + Send + Sync>> {
+        let binary_path = rustfs_binary_path();
+        self.start_node_from_binary(node_idx, &binary_path).await
+    }
+
+    /// Start one stopped cluster node with a specific RustFS binary while
+    /// preserving the cluster's volume layout and that node's data directory.
+    pub async fn start_node_from_binary(
+        &mut self,
+        node_idx: usize,
+        binary_path: &Path,
+    ) -> Result<(), Box<dyn std::error::Error + Send + Sync>> {
+        let volumes_arg = self.build_volumes_arg();
+        self.spawn_node(node_idx, binary_path, &volumes_arg)?;
+
+        self.wait_for_node_ready(&self.nodes[node_idx].address, node_idx).await?;
+        self.wait_for_node_service_ready(node_idx).await?;
+        Ok(())
+    }
+
+    fn spawn_node(
+        &mut self,
+        node_idx: usize,
+        binary_path: &Path,
+        volumes_arg: &str,
+    ) -> Result<(), Box<dyn std::error::Error + Send + Sync>> {
        self.ensure_node_index(node_idx)?;
        if self.nodes[node_idx].process.is_some() {
            return Err(format!("cluster node {node_idx} is already running").into());
        }
+        if !binary_path.is_file() {
+            return Err(format!("RustFS binary does not exist: {}", binary_path.display()).into());
+        }

-        let binary_path = rustfs_binary_path();
-        let volumes_arg = self.build_volumes_arg();
        let log_path = self.node_capture_log_paths[node_idx].clone();
        let node = &mut self.nodes[node_idx];
-        info!("Starting cluster node {} on {}", node_idx, node.address);
+        info!("Starting cluster node {} on {} with {}", node_idx, node.address, binary_path.display());

-        let mut command = Command::new(&binary_path);
+        let mut command = Command::new(binary_path);
        command
-            .env("RUSTFS_VOLUMES", &volumes_arg)
+            .env("RUSTFS_VOLUMES", volumes_arg)
            .env("RUSTFS_ADDRESS", &node.address)
            .env("RUSTFS_ACCESS_KEY", &self.access_key)
            .env("RUSTFS_SECRET_KEY", &self.secret_key)
@@ -1423,9 +1552,6 @@ impl RustFSTestClusterEnvironment {

        let process = command.current_dir(&node.data_dir).spawn()?;
        node.process = Some(process);
-
-        self.wait_for_node_ready(&self.nodes[node_idx].address, node_idx).await?;
-        self.wait_for_node_service_ready(node_idx).await?;
        Ok(())
    }

@@ -1573,6 +1699,51 @@ impl RustFSTestClusterEnvironment {
        process.wait()?;
        Ok(())
    }
+
+    /// Gracefully stop one cluster node and wait for its process to exit.
+    ///
+    /// This is intentionally separate from [`Self::stop_node`]: the latter is
+    /// a hard kill used by crash-recovery tests, while this path lets RustFS
+    /// complete its normal shutdown hooks before a test restarts the node.
+    pub async fn stop_node_gracefully(&mut self, node_idx: usize) -> Result<(), Box<dyn std::error::Error + Send + Sync>> {
+        self.ensure_node_index(node_idx)?;
+
+        #[cfg(unix)]
+        {
+            let Some(process) = self.nodes[node_idx].process.as_ref() else {
+                return Ok(());
+            };
+            let pid = process.id().to_string();
+            let signal_status = Command::new("kill").args(["-TERM", &pid]).status()?;
+            if !signal_status.success() {
+                return Err(format!("failed to send SIGTERM to cluster node {node_idx} (pid {pid})").into());
+            }
+
+            let mut process = self.nodes[node_idx]
+                .process
+                .take()
+                .ok_or_else(|| format!("cluster node {node_idx} process disappeared while stopping"))?;
+            let deadline = std::time::Instant::now() + Duration::from_secs(45);
+            loop {
+                if let Some(status) = process.try_wait()? {
+                    info!("Cluster node {} stopped gracefully with {}", node_idx, status);
+                    return Ok(());
+                }
+                if std::time::Instant::now() >= deadline {
+                    let _ = process.kill();
+                    let _ = process.wait();
+                    return Err(format!("cluster node {node_idx} did not stop gracefully within 45 seconds").into());
+                }
+                sleep(Duration::from_millis(100)).await;
+            }
+        }
+
+        #[cfg(not(unix))]
+        {
+            let _ = node_idx;
+            Err("graceful cluster-node stop is only supported on Unix E2E hosts".into())
+        }
+    }
 }

 impl Drop for RustFSTestClusterEnvironment {
@@ -1715,30 +1886,128 @@ pub(crate) async fn admin_create_user(
    username: &str,
    secret_key: &str,
 ) -> Result<(), Box<dyn std::error::Error + Send + Sync>> {
-    let url = format!("{}/rustfs/admin/v3/add-user?accessKey={}", env.url, username);
-    let body = serde_json::json!({
-        "secretKey": secret_key,
-        "status": "enabled"
-    });
-    let response = signed_request(
-        http::Method::PUT,
-        &url,
-        &env.access_key,
-        &env.secret_key,
-        Some(body.to_string().into_bytes()),
-        Some("application/json"),
-    )
-    .await?;
+    admin_create_user_via(AdminTransport::Signed, &env.url, &env.access_key, &env.secret_key, username, secret_key).await
+}

-    if response.status() != reqwest::StatusCode::OK {
-        let status = response.status();
-        let body = response.text().await.unwrap_or_default();
-        return Err(format!("create user failed: {status} {body}").into());
+/// Transport used by the shared admin-API helpers: in-process SigV4 signing
+/// via [`signed_request`], or the external `awscurl` binary (an independent
+/// SigV4 implementation exercised by the awscurl-gated suites).
+#[derive(Clone, Copy, Debug, PartialEq, Eq)]
+pub(crate) enum AdminTransport {
+    Signed,
+    Awscurl,
+}
+
+/// Execute an admin-API request against `base_url` with admin credentials over
+/// the chosen transport, failing on any non-success response.
+pub(crate) async fn admin_execute_at(
+    transport: AdminTransport,
+    method: http::Method,
+    base_url: &str,
+    admin_access_key: &str,
+    admin_secret_key: &str,
+    path_and_query: &str,
+    body: Option<&str>,
+) -> Result<(), Box<dyn std::error::Error + Send + Sync>> {
+    let url = format!("{base_url}{path_and_query}");
+    match transport {
+        AdminTransport::Signed => {
+            let content_type = match body {
+                Some(body) if !body.is_empty() => Some("application/json"),
+                _ => None,
+            };
+            let response = signed_request(
+                method.clone(),
+                &url,
+                admin_access_key,
+                admin_secret_key,
+                body.map(|body| body.as_bytes().to_vec()),
+                content_type,
+            )
+            .await?;
+            if !response.status().is_success() {
+                let status = response.status();
+                let text = response.text().await.unwrap_or_default();
+                return Err(format!("{method} {path_and_query} failed: {status} {text}").into());
+            }
+        }
+        AdminTransport::Awscurl => {
+            execute_awscurl(&url, method.as_str(), body, admin_access_key, admin_secret_key).await?;
+        }
    }
-
    Ok(())
 }

+/// Create a new IAM user via the admin API over the chosen transport.
+pub(crate) async fn admin_create_user_via(
+    transport: AdminTransport,
+    base_url: &str,
+    admin_access_key: &str,
+    admin_secret_key: &str,
+    username: &str,
+    secret_key: &str,
+) -> Result<(), Box<dyn std::error::Error + Send + Sync>> {
+    let path = format!("/rustfs/admin/v3/add-user?accessKey={username}");
+    let body = serde_json::json!({"secretKey": secret_key, "status": "enabled"}).to_string();
+    admin_execute_at(
+        transport,
+        http::Method::PUT,
+        base_url,
+        admin_access_key,
+        admin_secret_key,
+        &path,
+        Some(&body),
+    )
+    .await
+}
+
+/// Install a canned policy via the admin API over the chosen transport.
+pub(crate) async fn admin_add_canned_policy_via(
+    transport: AdminTransport,
+    base_url: &str,
+    admin_access_key: &str,
+    admin_secret_key: &str,
+    policy_name: &str,
+    policy_json: &str,
+) -> Result<(), Box<dyn std::error::Error + Send + Sync>> {
+    let path = format!("/rustfs/admin/v3/add-canned-policy?name={policy_name}");
+    admin_execute_at(
+        transport,
+        http::Method::PUT,
+        base_url,
+        admin_access_key,
+        admin_secret_key,
+        &path,
+        Some(policy_json),
+    )
+    .await
+}
+
+/// Attach a canned policy to a user via the admin API over the chosen transport.
+pub(crate) async fn admin_attach_user_policy_via(
+    transport: AdminTransport,
+    base_url: &str,
+    admin_access_key: &str,
+    admin_secret_key: &str,
+    policy_name: &str,
+    username: &str,
+) -> Result<(), Box<dyn std::error::Error + Send + Sync>> {
+    let path = format!("/rustfs/admin/v3/set-user-or-group-policy?policyName={policy_name}&userOrGroup={username}&isGroup=false");
+    // `Some("")` preserves the historical wire shape on both transports: awscurl
+    // keeps sending `-d ''` and the signed path attaches an empty body with no
+    // content type.
+    admin_execute_at(
+        transport,
+        http::Method::PUT,
+        base_url,
+        admin_access_key,
+        admin_secret_key,
+        &path,
+        Some(""),
+    )
+    .await
+}
+
 #[cfg(test)]
 mod tests {
    use super::*;
@@ -1776,6 +2045,27 @@ mod tests {
        );
    }

+    #[test]
+    fn resolves_rustfs_binary_in_configured_cargo_target_directory() {
+        let workspace = Path::new("workspace");
+        let profile = if cfg!(debug_assertions) { "debug" } else { "release" };
+        let binary = format!("rustfs{}", std::env::consts::EXE_SUFFIX);
+        assert_eq!(
+            resolve_rustfs_binary_path(workspace, None),
+            workspace.join("target").join(profile).join(&binary)
+        );
+        assert_eq!(
+            resolve_rustfs_binary_path(workspace, Some(Path::new("custom-target"))),
+            workspace.join("custom-target").join(profile).join(&binary)
+        );
+
+        let absolute = std::env::temp_dir().join("rustfs-e2e-custom-target");
+        assert_eq!(
+            resolve_rustfs_binary_path(workspace, Some(&absolute)),
+            absolute.join(profile).join(binary)
+        );
+    }
+
    #[test]
    fn full_feature_enables_any_required_feature() {
        assert!(rustfs_build_feature_enabled(Some("full"), "sftp"));
@@ -1809,7 +2099,7 @@ mod tests {
        }
        let multidrive = topology.drives_per_node > 1;

-        let nodes = (0..topology.node_count)
+        let nodes: Vec<ClusterNode> = (0..topology.node_count)
            .map(|i| {
                let address = format!("127.0.0.1:{}", 9000 + i);
                let data_dirs: Vec<String> = if multidrive {
@@ -1830,6 +2120,7 @@ mod tests {
            })
            .collect();

+        let node_count = nodes.len();
        RustFSTestClusterEnvironment {
            nodes,
            temp_dir,
@@ -1839,6 +2130,7 @@ mod tests {
            node_extra_env: vec![Vec::new(); topology.node_count],
            node_capture_log_paths: vec![None; topology.node_count],
            topology,
+            volume_proxy_addresses: vec![None; node_count],
        }
    }

@@ -1923,6 +2215,25 @@ mod tests {
        assert!(ClusterTopology::single_pool_multidrive(1, 1).validate().is_ok());
    }

+    #[tokio::test]
+    async fn volume_proxy_rewrites_cluster_volume_endpoint() {
+        let mut env = RustFSTestClusterEnvironment::new(1)
+            .await
+            .expect("cluster environment should allocate a node");
+        let direct = env.nodes[0].address.clone();
+        let proxy = env
+            .start_volume_proxy_for_node(0)
+            .await
+            .expect("volume proxy should bind before the target server starts");
+        let proxied = proxy.local_addr().to_string();
+        let volumes = env.rustfs_volumes_arg();
+
+        assert!(volumes.contains(&proxied), "volumes must use the proxy address: {volumes}");
+        assert!(!volumes.contains(&direct), "volumes must not retain the direct address: {volumes}");
+
+        proxy.shutdown().await;
+    }
+
    #[test]
    fn cluster_node_env_supports_per_node_overrides() {
        let mut env = fake_cluster(ClusterTopology::single_pool(4));
@@ -4,7 +4,8 @@ use crate::common::{RustFSTestEnvironment, init_logging, rustfs_binary_path};
 use aws_sdk_s3::primitives::ByteStream;
 use aws_sdk_s3::types::{CompletedMultipartUpload, CompletedPart};
 use std::fs;
-use std::path::PathBuf;
+use std::io;
+use std::path::{Path, PathBuf};
 use std::process::Command;
 use tracing::info;

@@ -31,30 +32,58 @@ fn generate_high_ratio_binary_data(size: usize, seed: u8) -> Vec<u8> {
        .collect()
 }

-fn find_part_files(temp_dir: &str, bucket: &str, object_key: &str) -> Vec<PathBuf> {
+fn find_part_files(temp_dir: &str, bucket: &str, object_key: &str) -> io::Result<Vec<PathBuf>> {
    let bucket_path = PathBuf::from(temp_dir).join(bucket);
    let mut part_files = Vec::new();

-    fn scan_dir(dir: &PathBuf, target: &str, results: &mut Vec<PathBuf>) {
-        if let Ok(entries) = fs::read_dir(dir) {
-            for entry in entries.flatten() {
-                let path = entry.path();
-                if path.is_dir() {
-                    scan_dir(&path, target, results);
-                } else if path
-                    .file_name()
-                    .map(|n| n.to_string_lossy().starts_with("part."))
-                    .unwrap_or(false)
-                    && path.to_string_lossy().contains(target)
-                {
-                    results.push(path);
+    fn scan_dir(dir: &Path, target: &str, results: &mut Vec<PathBuf>) -> io::Result<()> {
+        let entries = fs::read_dir(dir)
+            .map_err(|error| io::Error::new(error.kind(), format!("failed to read {}: {error}", dir.display())))?;
+        for entry in entries {
+            let entry = entry
+                .map_err(|error| io::Error::new(error.kind(), format!("failed to read entry in {}: {error}", dir.display())))?;
+            let path = entry.path();
+            let file_type = entry
+                .file_type()
+                .map_err(|error| io::Error::new(error.kind(), format!("failed to inspect {}: {error}", path.display())))?;
+            if file_type.is_dir() {
+                scan_dir(&path, target, results)?;
+            } else if path
+                .file_name()
+                .map(|n| n.to_string_lossy().starts_with("part."))
+                .unwrap_or(false)
+                && path.to_string_lossy().contains(target)
+            {
+                if !file_type.is_file() {
+                    return Err(io::Error::new(
+                        io::ErrorKind::InvalidData,
+                        format!("expected regular part file at {}", path.display()),
+                    ));
                }
+                results.push(path);
            }
        }
+        Ok(())
    }

-    scan_dir(&bucket_path, object_key, &mut part_files);
-    part_files
+    scan_dir(&bucket_path, object_key, &mut part_files)?;
+    Ok(part_files)
+}
+
+fn part_files_total_size(part_files: &[PathBuf]) -> io::Result<u64> {
+    part_files.iter().try_fold(0_u64, |total, path| {
+        let metadata = fs::symlink_metadata(path)
+            .map_err(|error| io::Error::new(error.kind(), format!("failed to stat {}: {error}", path.display())))?;
+        if !metadata.file_type().is_file() {
+            return Err(io::Error::new(
+                io::ErrorKind::InvalidData,
+                format!("expected regular part file at {}", path.display()),
+            ));
+        }
+        total
+            .checked_add(metadata.len())
+            .ok_or_else(|| io::Error::new(io::ErrorKind::InvalidData, "on-disk part size overflow"))
+    })
 }

 async fn start_rustfs_with_compression(env: &mut RustFSTestEnvironment) -> Result<(), Box<dyn std::error::Error + Send + Sync>> {
@@ -123,8 +152,9 @@ async fn test_compression_roundtrip() -> Result<(), Box<dyn std::error::Error +
    let content_length = head_response.content_length().unwrap_or(0);
    assert_eq!(content_length as usize, original_size, "Content-Length should be original size");

-    let part_files = find_part_files(&env.temp_dir, COMPRESSION_TEST_BUCKET, object_key);
-    let total_physical_size: u64 = part_files.iter().filter_map(|p| fs::metadata(p).ok()).map(|m| m.len()).sum();
+    let part_files = find_part_files(&env.temp_dir, COMPRESSION_TEST_BUCKET, object_key)?;
+    assert!(!part_files.is_empty(), "expected on-disk part files for the compressed object");
+    let total_physical_size = part_files_total_size(&part_files)?;

    assert!(
        total_physical_size < original_size as u64,
@@ -246,9 +276,9 @@ async fn test_compression_multipart_roundtrip() -> Result<(), Box<dyn std::error
        "Content-Length should be the logical object size"
    );

-    let part_files = find_part_files(&env.temp_dir, MULTIPART_COMPRESSION_BUCKET, object_key);
+    let part_files = find_part_files(&env.temp_dir, MULTIPART_COMPRESSION_BUCKET, object_key)?;
    assert!(!part_files.is_empty(), "expected on-disk part files for the multipart object");
-    let total_physical_size: u64 = part_files.iter().filter_map(|p| fs::metadata(p).ok()).map(|m| m.len()).sum();
+    let total_physical_size = part_files_total_size(&part_files)?;
    assert!(
        total_physical_size < (total_size / 2) as u64,
        "Physical size {total_physical_size} should be well below original size {total_size} (multipart compression applied)"
@@ -366,9 +396,9 @@ async fn test_compression_multipart_high_ratio_binary_roundtrip() -> Result<(),

    // This pattern compresses to roughly 1/50 of its logical size, so a comfortably loose 2x
    // margin still proves the parts were stored compressed rather than raw or double-encoded.
-    let part_files = find_part_files(&env.temp_dir, MPU_HIGH_RATIO_BUCKET, object_key);
+    let part_files = find_part_files(&env.temp_dir, MPU_HIGH_RATIO_BUCKET, object_key)?;
    assert!(!part_files.is_empty(), "expected on-disk part files for the multipart object");
-    let total_physical_size: u64 = part_files.iter().filter_map(|p| fs::metadata(p).ok()).map(|m| m.len()).sum();
+    let total_physical_size = part_files_total_size(&part_files)?;
    assert!(
        total_physical_size < (total_size as u64) / 2,
        "Physical size {total_physical_size} should be far below the logical size {total_size} for high-ratio data"
@@ -522,9 +552,9 @@ async fn test_compression_multipart_upload_part_copy_roundtrip() -> Result<(), B
        "Content-Length should be the logical object size"
    );

-    let part_files = find_part_files(&env.temp_dir, MPU_COPY_COMPRESSION_BUCKET, target_key);
+    let part_files = find_part_files(&env.temp_dir, MPU_COPY_COMPRESSION_BUCKET, target_key)?;
    assert!(!part_files.is_empty(), "expected on-disk part files for the copied object");
-    let total_physical_size: u64 = part_files.iter().filter_map(|p| fs::metadata(p).ok()).map(|m| m.len()).sum();
+    let total_physical_size = part_files_total_size(&part_files)?;
    assert!(
        total_physical_size < (total_size / 2) as u64,
        "Physical size {total_physical_size} should be well below original size {total_size} (copied part compression applied)"
@@ -585,9 +615,9 @@ async fn test_compression_multipart_three_parts_part_number_gets() -> Result<(),
        "Content-Length should be the logical object size"
    );

-    let part_files = find_part_files(&env.temp_dir, MPU_THREE_PARTS_BUCKET, object_key);
+    let part_files = find_part_files(&env.temp_dir, MPU_THREE_PARTS_BUCKET, object_key)?;
    assert!(!part_files.is_empty(), "expected on-disk part files for the multipart object");
-    let total_physical_size: u64 = part_files.iter().filter_map(|p| fs::metadata(p).ok()).map(|m| m.len()).sum();
+    let total_physical_size = part_files_total_size(&part_files)?;
    assert!(
        total_physical_size < (total_size / 2) as u64,
        "Physical size {total_physical_size} should be well below original size {total_size} (multipart compression applied)"
@@ -622,11 +652,10 @@ const MPU_SSE_COMPRESSION_BUCKET: &str = "compression-mpu-sse-bucket";
 async fn start_rustfs_with_compression_and_sse(
    env: &mut RustFSTestEnvironment,
 ) -> Result<(), Box<dyn std::error::Error + Send + Sync>> {
-    use base64::Engine;
    env.cleanup_existing_processes().await?;

    let binary_path = rustfs_binary_path();
-    let master_key = base64::engine::general_purpose::STANDARD.encode([0x42u8; 32]);
+    let master_key = base64_simd::STANDARD.encode_to_string([0x42u8; 32]);
    // Server output goes to a file inside the per-test temp dir so a failing
    // run can be diagnosed from the child's logs.
    let server_log = std::fs::File::create(format!("{}/server.log", env.temp_dir))?;
@@ -734,9 +763,9 @@ async fn test_compression_multipart_sse_s3_roundtrip() -> Result<(), Box<dyn std
        "HEAD must report SSE-S3"
    );

-    let part_files = find_part_files(&env.temp_dir, MPU_SSE_COMPRESSION_BUCKET, object_key);
+    let part_files = find_part_files(&env.temp_dir, MPU_SSE_COMPRESSION_BUCKET, object_key)?;
    assert!(!part_files.is_empty(), "expected on-disk part files for the multipart object");
-    let total_physical_size: u64 = part_files.iter().filter_map(|p| fs::metadata(p).ok()).map(|m| m.len()).sum();
+    let total_physical_size = part_files_total_size(&part_files)?;
    assert!(
        total_physical_size < (total_size / 2) as u64,
        "Physical size {total_physical_size} should be well below original size {total_size} (compress-then-encrypt applied)"
@@ -27,8 +27,7 @@ mod tests {
        VersioningConfiguration,
    };
    use aws_smithy_http_client::Builder as SmithyHttpClientBuilder;
-    use base64::Engine as _;
-    use base64::engine::general_purpose::STANDARD as BASE64;
+    use base64_simd::STANDARD as BASE64;
    use rustfs_rio::{Checksum, ChecksumType as RioChecksumType};
    use sha2::{Digest, Sha256};
    use tracing::info;
@@ -465,7 +464,7 @@ mod tests {
        create_versioned_bucket(&client, dst_bucket).await;

        let content = b"deterministic synthetic payload for copy-object checksum #4996";
-        let expected_sha256 = BASE64.encode(Sha256::digest(content));
+        let expected_sha256 = BASE64.encode_to_string(Sha256::digest(content));

        client
            .put_object()
@@ -534,7 +533,7 @@ mod tests {
        create_versioned_bucket(&client, dst_bucket).await;

        let content = b"another deterministic payload whose source checksum must survive the copy";
-        let expected_sha256 = BASE64.encode(Sha256::digest(content));
+        let expected_sha256 = BASE64.encode_to_string(Sha256::digest(content));

        // Store the source WITH a SHA-256 checksum so it has one to preserve.
        let put_src = client
@@ -614,7 +613,7 @@ mod tests {
        create_versioned_bucket(&client, dst_bucket).await;

        let content = b"payload whose copy must be re-checksummed with a different algorithm";
-        let expected_sha256 = BASE64.encode(Sha256::digest(content));
+        let expected_sha256 = BASE64.encode_to_string(Sha256::digest(content));

        // Source is stored WITH a SHA-256 checksum.
        client
@@ -59,7 +59,6 @@ where
 /// Regression test for data usage accuracy (issue #1012).
 /// Launches rustfs, writes 1000 objects, then asserts admin data usage reports the full count.
 #[tokio::test(flavor = "multi_thread")]
-#[ignore = "Starts a rustfs server and requires awscurl; enable when running full E2E"]
 async fn data_usage_reports_all_objects() -> Result<(), Box<dyn std::error::Error + Send + Sync>> {
    init_logging();

@@ -86,28 +85,20 @@ async fn data_usage_reports_all_objects() -> Result<(), Box<dyn std::error::Erro
        usage
            .buckets_usage
            .get(TEST_BUCKET)
-            .map(|bucket_usage| usage.objects_total_count >= 1000 && bucket_usage.objects_count >= 1000)
+            .map(|bucket_usage| usage.objects_total_count == 1000 && bucket_usage.objects_count == 1000)
            .unwrap_or(false)
    })
    .await?;

-    // Assert total object count and per-bucket count are not truncated
+    // Assert total object count and per-bucket count are exact.
    let bucket_usage = usage
        .buckets_usage
        .get(TEST_BUCKET)
        .cloned()
        .expect("bucket usage should exist");

-    assert!(
-        usage.objects_total_count >= 1000,
-        "total object count should be at least 1000, got {}",
-        usage.objects_total_count
-    );
-    assert!(
-        bucket_usage.objects_count >= 1000,
-        "bucket object count should be at least 1000, got {}",
-        bucket_usage.objects_count
-    );
+    assert_eq!(usage.objects_total_count, 1000, "total object count should be exact");
+    assert_eq!(bucket_usage.objects_count, 1000, "bucket object count should be exact");

    env.stop_server();
    Ok(())
@@ -116,7 +107,6 @@ async fn data_usage_reports_all_objects() -> Result<(), Box<dyn std::error::Erro
 /// Regression test for issue #3898.
 /// Versioned buckets should expose versions and delete markers through admin data usage.
 #[tokio::test(flavor = "multi_thread")]
-#[ignore = "Starts a rustfs server and requires awscurl; enable when running full E2E"]
 async fn data_usage_reports_versioned_objects_and_delete_markers() -> Result<(), Box<dyn std::error::Error + Send + Sync>> {
    init_logging();

@@ -0,0 +1,174 @@
+// Copyright 2024 RustFS Team
+//
+// Licensed under the Apache License, Version 2.0 (the "License");
+// you may not use this file except in compliance with the License.
+// You may obtain a copy of the License at
+//
+//     http://www.apache.org/licenses/LICENSE-2.0
+//
+// Unless required by applicable law or agreed to in writing, software
+// distributed under the License is distributed on an "AS IS" BASIS,
+// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+// See the License for the specific language governing permissions and
+// limitations under the License.
+
+//! Regression: an object legally committed at degraded write quorum must stay
+//! listable while a *different* drive is offline.
+//!
+//! On a 4-drive EC 2+2 set, a PUT made while one drive is down persists
+//! `xl.meta` on 3 of 4 drives (write quorum). If a different drive later goes
+//! offline before heal converges, a strict latest-listing quorum of 3 can only
+//! ever observe 2 copies, so ListObjectsV2 silently dropped the object even
+//! though GetObject (read quorum 2) still succeeded. Exposed by the flaky
+//! "Mixed-version rolling upgrade from rc.2" CI lane (run 33478999853); the
+//! product fix relaxes the listing's required object quorum by the number of
+//! set drives the listing could not consult (see
+//! `latest_listing_required_object_quorum` in
+//! `crates/ecstore/src/store/list_objects.rs`).
+
+#[cfg(test)]
+mod tests {
+    use crate::common::{RustFSTestClusterEnvironment, init_logging};
+    use aws_sdk_s3::Client;
+    use bytes::Bytes;
+    use std::collections::HashSet;
+    use std::error::Error;
+    use std::time::{Duration, Instant};
+    use tracing::info;
+
+    type TestResult = Result<(), Box<dyn Error + Send + Sync>>;
+
+    const BUCKET: &str = "degraded-listing-availability";
+    const OBJECT_COUNT: usize = 8;
+    /// Well under the observed heal-convergence gap (~50s in the CI incident),
+    /// so a listing that only completes after heal restores the missing copy
+    /// still fails this deadline on a regressed build.
+    const LISTING_DEADLINE: Duration = Duration::from_secs(25);
+    const GET_RETRY_DEADLINE: Duration = Duration::from_secs(15);
+    const PUT_RETRY_DEADLINE: Duration = Duration::from_secs(15);
+
+    fn object_key(idx: usize) -> String {
+        format!("degraded-object-{idx:02}")
+    }
+
+    async fn list_all_keys(client: &Client) -> Result<HashSet<String>, Box<dyn Error + Send + Sync>> {
+        let mut keys = HashSet::new();
+        let mut continuation_token: Option<String> = None;
+        loop {
+            let response = client
+                .list_objects_v2()
+                .bucket(BUCKET)
+                .set_continuation_token(continuation_token.clone())
+                .send()
+                .await?;
+            keys.extend(
+                response
+                    .contents()
+                    .iter()
+                    .filter_map(|object| object.key().map(str::to_owned)),
+            );
+            match response.next_continuation_token() {
+                Some(token) => continuation_token = Some(token.to_owned()),
+                None => break,
+            }
+        }
+        Ok(keys)
+    }
+
+    /// 4-node single-drive cluster (EC 2+2, write quorum 3):
+    /// 1. Stop node 1 and PUT objects — each commits on nodes {0, 2, 3} only.
+    /// 2. Stop node 3 (a holder drive), then bring node 1 back before heal can
+    ///    recreate the missing copies there.
+    /// 3. Every object still satisfies read quorum (nodes 0 and 2), so GET
+    ///    must succeed AND ListObjectsV2 must report every key well before
+    ///    heal converges.
+    #[tokio::test]
+    async fn degraded_write_remains_listable_while_a_different_drive_is_offline() -> TestResult {
+        init_logging();
+
+        let mut cluster = RustFSTestClusterEnvironment::new(4).await?;
+        // Listing availability must not depend on heal convergence: disable
+        // the background healers so the degraded objects keep their metadata
+        // on exactly 3 of 4 drives for the whole test.
+        cluster.set_env("RUSTFS_HEAL_ENABLED", "false");
+        cluster.set_env("RUSTFS_SCANNER_ENABLED", "false");
+        cluster.start().await?;
+        cluster.create_test_bucket(BUCKET).await?;
+        let client = cluster.create_s3_client(0)?;
+
+        info!("stopping node 1 so the uploads commit at degraded write quorum (3 of 4)");
+        cluster.stop_node(1)?;
+        // The first writes after a node drops can see transient 503s while the
+        // survivors notice the dead peer; retry briefly (overwrites of the same
+        // unversioned key are idempotent).
+        for idx in 0..OBJECT_COUNT {
+            let key = object_key(idx);
+            let body = format!("degraded listing payload {idx}");
+            let deadline = Instant::now() + PUT_RETRY_DEADLINE;
+            loop {
+                let request = client
+                    .put_object()
+                    .bucket(BUCKET)
+                    .key(&key)
+                    .body(Bytes::from(body.clone()).into());
+                match request.send().await {
+                    Ok(_) => break,
+                    Err(error) if Instant::now() < deadline => {
+                        info!("retrying degraded PUT for {key}: {error}");
+                        tokio::time::sleep(Duration::from_millis(500)).await;
+                    }
+                    Err(error) => return Err(format!("degraded PUT for {key} failed: {error}").into()),
+                }
+            }
+        }
+
+        info!("stopping node 3 (holds a copy) and restoring node 1 (holds none)");
+        cluster.stop_node(3)?;
+        cluster.start_node(1).await?;
+
+        // The first requests after a node drops can see transient 503s while
+        // the survivors notice the dead peer; retry briefly before asserting.
+        for idx in 0..OBJECT_COUNT {
+            let key = object_key(idx);
+            let deadline = Instant::now() + GET_RETRY_DEADLINE;
+            let body = loop {
+                match client.get_object().bucket(BUCKET).key(&key).send().await {
+                    Ok(response) => break response.body.collect().await?.into_bytes(),
+                    Err(error) if Instant::now() < deadline => {
+                        info!("retrying degraded GET for {key}: {error}");
+                        tokio::time::sleep(Duration::from_millis(500)).await;
+                    }
+                    Err(error) => return Err(format!("degraded object {key} failed read quorum GET: {error}").into()),
+                }
+            };
+            assert!(!body.is_empty(), "degraded object {key} should read back at read quorum");
+        }
+
+        let expected: HashSet<String> = (0..OBJECT_COUNT).map(object_key).collect();
+        let deadline = Instant::now() + LISTING_DEADLINE;
+        let listed = loop {
+            let listed = match list_all_keys(&client).await {
+                Ok(keys) => keys,
+                Err(error) if Instant::now() < deadline => {
+                    info!("retrying degraded listing: {error}");
+                    tokio::time::sleep(Duration::from_millis(500)).await;
+                    continue;
+                }
+                Err(error) => return Err(error),
+            };
+            if expected.is_subset(&listed) {
+                break listed;
+            }
+            assert!(
+                Instant::now() < deadline,
+                "objects readable at read quorum stayed missing from ListObjectsV2 for {LISTING_DEADLINE:?}: \
+                 missing={:?} listed={listed:?}",
+                expected.difference(&listed).collect::<Vec<_>>(),
+            );
+            tokio::time::sleep(Duration::from_millis(500)).await;
+        };
+        info!(listed = listed.len(), "degraded objects are listable while node 3 is offline");
+
+        Ok(())
+    }
+}
@@ -23,8 +23,8 @@
 //! It was fixed in three layers on `main`, each with its own *unit* regression:
 //!   * rustfs#4594 — `GetObjectStreamingReader::poll_read` now returns
 //!     `UnexpectedEof` on a short body instead of a clean `Ok(())`
-//!     (`rustfs/src/app/object_usecase.rs`,
-//!     `app::object_usecase::tests::get_object_streaming_reader_errors_on_short_eof`).
+//!     (`rustfs/src/app/object/get.rs`,
+//!     `app::object::get::tests::get_object_streaming_reader_errors_on_short_eof`).
 //!   * rustfs#4560 — the lazy multipart codec reader degrades a later part to
 //!     the legacy per-part decode in place, and surfaces reconstruction errors
 //!     instead of silently truncating
@@ -155,8 +155,13 @@ mod tests {
            .key(key)
            .version_id(version_id)
            .send()
-            .await;
-        assert!(get_deleted_version.is_err(), "explicitly deleted version should no longer be readable");
+            .await
+            .expect_err("explicitly deleted version should no longer be readable");
+        assert_eq!(
+            get_deleted_version.raw_response().map(|response| response.status().as_u16()),
+            Some(404),
+            "explicitly deleted version absence probe must return HTTP 404, got {get_deleted_version:?}"
+        );

        Ok(())
    }
@@ -117,9 +117,18 @@ mod tests {
        );

        // Verify HEAD returns 404
-        let head = client.head_object().bucket(bucket).key("to-delete.txt").send().await;
-
-        assert!(head.is_err(), "RT-05 FAIL: HEAD on deleted object should return error, got success");
+        let error = client
+            .head_object()
+            .bucket(bucket)
+            .key("to-delete.txt")
+            .send()
+            .await
+            .expect_err("RT-05 FAIL: HEAD on deleted object should return 404, got success");
+        assert_eq!(
+            error.raw_response().map(|response| response.status().as_u16()),
+            Some(404),
+            "RT-05 FAIL: HEAD on deleted object must return HTTP 404, got {error:?}"
+        );

        info!("RT-05 PASS: delete correctly removes object from LIST and HEAD");
        Ok(())
@@ -414,9 +423,18 @@ mod tests {

        // All HEAD requests should return 404
        for key in &keys {
-            let head = client.head_object().bucket(bucket).key(*key).send().await;
-
-            assert!(head.is_err(), "RT-05f FAIL: HEAD on deleted key '{key}' should return error");
+            let error = client
+                .head_object()
+                .bucket(bucket)
+                .key(*key)
+                .send()
+                .await
+                .expect_err("RT-05f FAIL: HEAD on deleted key should return 404, got success");
+            assert_eq!(
+                error.raw_response().map(|response| response.status().as_u16()),
+                Some(404),
+                "RT-05f FAIL: HEAD on deleted key '{key}' must return HTTP 404, got {error:?}"
+            );
        }

        // LIST should be empty
@@ -16,36 +16,29 @@
 //! session policy** (`Policy` parameter) via `awscurl --service sts` with explicit
 //! `Content-Type: application/x-www-form-urlencoded` on `POST /`.

-use crate::common::{RustFSTestEnvironment, awscurl_delete, awscurl_post_sts_form_urlencoded, awscurl_put, init_logging};
-use aws_sdk_s3::config::{Credentials, Region};
+use crate::common::{
+    AdminTransport, RustFSTestEnvironment, admin_add_canned_policy_via, admin_attach_user_policy_via, admin_create_user_via,
+    awscurl_delete, awscurl_post_sts_form_urlencoded, build_test_s3_config, init_logging,
+};
+use aws_sdk_s3::Client;
+use aws_sdk_s3::error::ProvideErrorMetadata;
 use aws_sdk_s3::primitives::ByteStream;
 use aws_sdk_s3::types::{Delete, ObjectIdentifier, Tag, Tagging};
-use aws_sdk_s3::{Client, Config};
 use tracing::info;
 use uuid::Uuid;

 fn user_client(env: &RustFSTestEnvironment, access_key: &str, secret_key: &str) -> Client {
-    let credentials = Credentials::new(access_key, secret_key, None, None, "e2e-existing-tag");
-    let config = Config::builder()
-        .credentials_provider(credentials)
-        .region(Region::new("us-east-1"))
-        .endpoint_url(&env.url)
-        .force_path_style(true)
-        .behavior_version_latest()
-        .build();
-    Client::from_conf(config)
+    env.create_s3_client_with_credentials(access_key, secret_key)
 }

 fn sts_session_client(env: &RustFSTestEnvironment, access_key: &str, secret_key: &str, session_token: &str) -> Client {
-    let credentials = Credentials::new(access_key, secret_key, Some(session_token.into()), None, "e2e-sts-session");
-    let config = Config::builder()
-        .credentials_provider(credentials)
-        .region(Region::new("us-east-1"))
-        .endpoint_url(&env.url)
-        .force_path_style(true)
-        .behavior_version_latest()
-        .build();
-    Client::from_conf(config)
+    Client::from_conf(build_test_s3_config(
+        &env.url,
+        access_key,
+        secret_key,
+        Some(session_token),
+        "e2e-sts-session",
+    ))
 }

 fn extract_xml_tag(xml: &str, tag: &str) -> Option<String> {
@@ -76,15 +69,16 @@ async fn assume_role_with_session_policy(
    parse_assume_role_credentials(&xml)
 }

+// This suite deliberately drives the admin API through the external `awscurl`
+// binary (an independent SigV4 implementation), so the wrappers below pin
+// `AdminTransport::Awscurl`.
+
 async fn admin_create_user(
    env: &RustFSTestEnvironment,
    username: &str,
    password: &str,
 ) -> Result<(), Box<dyn std::error::Error + Send + Sync>> {
-    let body = serde_json::json!({ "secretKey": password, "status": "enabled" }).to_string();
-    let url = format!("{}/rustfs/admin/v3/add-user?accessKey={}", env.url, username);
-    awscurl_put(&url, &body, &env.access_key, &env.secret_key).await?;
-    Ok(())
+    admin_create_user_via(AdminTransport::Awscurl, &env.url, &env.access_key, &env.secret_key, username, password).await
 }

 async fn admin_add_canned_policy(
@@ -92,9 +86,15 @@ async fn admin_add_canned_policy(
    policy_name: &str,
    policy_json: &str,
 ) -> Result<(), Box<dyn std::error::Error + Send + Sync>> {
-    let url = format!("{}/rustfs/admin/v3/add-canned-policy?name={}", env.url, policy_name);
-    awscurl_put(&url, policy_json, &env.access_key, &env.secret_key).await?;
-    Ok(())
+    admin_add_canned_policy_via(
+        AdminTransport::Awscurl,
+        &env.url,
+        &env.access_key,
+        &env.secret_key,
+        policy_name,
+        policy_json,
+    )
+    .await
 }

 async fn admin_attach_policy_to_user(
@@ -102,12 +102,7 @@ async fn admin_attach_policy_to_user(
    policy_name: &str,
    username: &str,
 ) -> Result<(), Box<dyn std::error::Error + Send + Sync>> {
-    let url = format!(
-        "{}/rustfs/admin/v3/set-user-or-group-policy?policyName={}&userOrGroup={}&isGroup=false",
-        env.url, policy_name, username
-    );
-    awscurl_put(&url, "", &env.access_key, &env.secret_key).await?;
-    Ok(())
+    admin_attach_user_policy_via(AdminTransport::Awscurl, &env.url, &env.access_key, &env.secret_key, policy_name, username).await
 }

 async fn admin_remove_user(env: &RustFSTestEnvironment, username: &str) {
@@ -208,10 +203,17 @@ async fn test_e2e_iam_policy_existing_object_tag_get_object() -> Result<(), Box<
    let _ = out.body.collect().await?;

    put_object_tag_kv(&admin, &bucket, key, "security", "private").await?;
-    let denied = uclient.get_object().bucket(&bucket).key(key).send().await;
-    assert!(
-        denied.is_err(),
-        "GetObject must be denied when ExistingObjectTag no longer matches IAM policy"
+    let denied = uclient
+        .get_object()
+        .bucket(&bucket)
+        .key(key)
+        .send()
+        .await
+        .expect_err("GetObject must be denied when ExistingObjectTag no longer matches IAM policy");
+    assert_eq!(
+        denied.as_service_error().and_then(ProvideErrorMetadata::code),
+        Some("AccessDenied"),
+        "IAM ExistingObjectTag mismatch must return AccessDenied: {denied:?}"
    );

    cleanup_bucket_and_object(&admin, &bucket, key).await;
@@ -245,8 +247,13 @@ async fn test_e2e_bucket_policy_existing_object_tag_get_object() -> Result<(), B
        .bucket(&bucket)
        .key(key)
        .send()
-        .await;
-    assert!(deny_before.is_err(), "without bucket policy, user must be denied");
+        .await
+        .expect_err("without bucket policy, user must be denied");
+    assert_eq!(
+        deny_before.as_service_error().and_then(ProvideErrorMetadata::code),
+        Some("AccessDenied"),
+        "missing bucket policy must return AccessDenied: {deny_before:?}"
+    );

    let bp = serde_json::json!({
        "Version": "2012-10-17",
@@ -268,8 +275,18 @@ async fn test_e2e_bucket_policy_existing_object_tag_get_object() -> Result<(), B
    let _ = ok.body.collect().await?;

    put_object_tag_kv(&admin, &bucket, key, "security", "private").await?;
-    let denied = uclient.get_object().bucket(&bucket).key(key).send().await;
-    assert!(denied.is_err(), "GetObject must fail when tag no longer satisfies bucket policy");
+    let denied = uclient
+        .get_object()
+        .bucket(&bucket)
+        .key(key)
+        .send()
+        .await
+        .expect_err("GetObject must fail when tag no longer satisfies bucket policy");
+    assert_eq!(
+        denied.as_service_error().and_then(ProvideErrorMetadata::code),
+        Some("AccessDenied"),
+        "bucket-policy ExistingObjectTag mismatch must return AccessDenied: {denied:?}"
+    );

    cleanup_bucket_and_object(&admin, &bucket, key).await;
    admin_remove_user(&env, &user).await;
@@ -335,10 +352,17 @@ async fn test_e2e_sts_assume_role_session_policy_existing_object_tag() -> Result
    let _ = ok.body.collect().await?;

    put_object_tag_kv(&parent_client, &bucket, key, "security", "private").await?;
-    let denied = session_client.get_object().bucket(&bucket).key(key).send().await;
-    assert!(
-        denied.is_err(),
-        "session policy must deny GetObject when ExistingObjectTag no longer matches"
+    let denied = session_client
+        .get_object()
+        .bucket(&bucket)
+        .key(key)
+        .send()
+        .await
+        .expect_err("session policy must deny GetObject when ExistingObjectTag no longer matches");
+    assert_eq!(
+        denied.as_service_error().and_then(ProvideErrorMetadata::code),
+        Some("AccessDenied"),
+        "STS ExistingObjectTag mismatch must return AccessDenied: {denied:?}"
    );

    cleanup_bucket_and_object(&admin, &bucket, key).await;
@@ -433,8 +457,18 @@ async fn test_e2e_sts_session_policy_delete_objects_object_prefix_only() -> Resu
    assert_eq!(error.key(), Some(denied_key));
    assert_eq!(error.code(), Some("AccessDenied"));

-    let allowed_head = parent_client.head_object().bucket(&bucket).key(allowed_key).send().await;
-    assert!(allowed_head.is_err(), "allowed-prefix object should have been deleted");
+    let allowed_head = parent_client
+        .head_object()
+        .bucket(&bucket)
+        .key(allowed_key)
+        .send()
+        .await
+        .expect_err("allowed-prefix object should have been deleted");
+    assert_eq!(
+        allowed_head.raw_response().map(|response| response.status().as_u16()),
+        Some(404),
+        "allowed-prefix object absence probe must return HTTP 404, got {allowed_head:?}"
+    );

    parent_client
        .head_object()
@@ -1,11 +1,15 @@
 # Programmable fake S3 target

-This module is the shared failure-injection boundary for replication end-to-end tests. It runs an in-process, path-style S3 endpoint backed by `s3s`; no production crate depends on it.
+This module is the shared failure-injection boundary for replication end-to-end tests and the programmable external source for on-demand-migration (ODM) tests. It runs an in-process, path-style S3 endpoint backed by `s3s`; no production crate depends on it.

 `FakeS3Target::start()` creates the listener. Add target buckets with `create_bucket`, point a RustFS remote target at `address()`, use `FAKE_ACCESS_KEY` / `FAKE_SECRET_KEY`, then enqueue per-operation faults with `inject`. Faults for one operation are consumed in FIFO order and do not consume faults queued for another operation. A fault is consumed only after `s3s` verifies the full request signature, so anonymous, other-access-key, and bad-signature traffic cannot disturb a script.

-Supported data operations are HeadBucket, GetBucketVersioning, PUT/GET/HEAD/DELETE Object, Get/Put/Delete ObjectTagging (tags live per version; Put replaces the whole set, Delete clears it), and create/upload/complete/abort multipart upload. `create_bucket` models general-purpose buckets in S3's shared global namespace; account-regional namespace buckets and their `-an` names are intentionally out of scope. Buckets are versioned: PUT creates a version, DELETE without `versionId` creates a delete marker, and DELETE with `versionId` removes exactly that version. Internal source version IDs must be UUIDs and are stored canonically. Source mtime is honored only for source-replication PUT/DELETE requests; absent or invalid values use receipt time, matching RustFS, while multipart completion always uses receipt time. Replicated versions are ordered newest-first by source mtime so late older versions and delete markers do not become current. Equal mtimes prefer objects over delete markers, then canonical UUID order; RustFS's internal FileMeta signature tie-break is intentionally out of scope because it is not part of the target S3 protocol. Multipart part numbers follow S3's `1..=10000` range, and every completed part except the final part must be at least 5 MiB.
+Supported data operations are HeadBucket, GetBucketVersioning, ListObjectsV2, PUT/GET/HEAD/DELETE Object, Get/Put/Delete ObjectTagging (tags live per version; Put replaces the whole set, Delete clears it), and create/upload/complete/abort multipart upload. `create_bucket` models general-purpose buckets in S3's shared global namespace; account-regional namespace buckets and their `-an` names are intentionally out of scope. Buckets created with `create_bucket` are versioned: PUT creates a version, DELETE without `versionId` creates a delete marker, and DELETE with `versionId` removes exactly that version. Internal source version IDs must be UUIDs and are stored canonically. Source mtime is honored only for source-replication PUT/DELETE requests; absent or invalid values use receipt time, matching RustFS, while multipart completion always uses receipt time. Replicated versions are ordered newest-first by source mtime so late older versions and delete markers do not become current. Equal mtimes prefer objects over delete markers, then canonical UUID order; RustFS's internal FileMeta signature tie-break is intentionally out of scope because it is not part of the target S3 protocol. Multipart part numbers follow S3's `1..=10000` range, and every completed part except the final part must be at least 5 MiB.

-Fault actions cover HTTP 401/403/503 responses, pre-dispatch delay, connection abort when a logical request-body threshold is reached, streaming slow drain, and a deliberately wrong response ETag (including multipart-complete XML). `requests()` returns the ordered, credential-free request journal for assertions. Each record also journals a `ProxyHeaderSnapshot` — the read-proxy anti-loop marker (`x-{rustfs,minio}-source-proxy-request`), the replication-check exemption header, and the client SSE-C header family (algorithm and key-MD5 values; for the key itself only its presence) — so proxy tests can pin the exact wire contract.
+`create_bucket_with_mode(name, BucketMode::Unversioned)` models a plain migration source: PUT overwrites in place, DELETE removes the key without a delete marker, GetBucketVersioning reports no status, and no `x-amz-version-id` is returned by PUT, GET, HEAD, tagging, or multipart completion. The only `versionId` such a bucket accepts is `null`; any other value is rejected with `InvalidArgument`. The mode is fixed at creation.

-The listener is loopback-only. It admits at most 64 active connections and two concurrently buffered request bodies; authenticated multipart-complete XML collection and assembly take both body permits. Keep-alive is disabled, request-header reads are bounded to 30 seconds, a parsed request is bounded to 65 seconds, and the complete connection lifetime is bounded to 100 seconds. It retains at most 256 buckets, 4,096 journal entries, 4,096 scripted faults, 4,096 object versions, 256 multipart uploads, and 10,000 multipart parts. Retained identifiers are capped at 1 KiB, user metadata at 2 KiB, and content type at 1 KiB. A PUT or uploaded part is capped at 64 MiB; a completed multipart object and all stored object/part data are capped at 128 MiB. Body drain, body-permit waits, delay, and slow-drain execution are bounded to 30 seconds; each slow-drain slice delay must be below that bound.
+ListObjectsV2 lists current versions only (a key whose newest version is a delete marker is hidden) in byte order and supports `prefix`, `delimiter`, `max-keys` (clamped to 1000), `start-after`, and `continuation-token`; common prefixes count toward `max-keys`, `IsTruncated` / `NextContinuationToken` / `KeyCount` follow S3, and continuation tokens are opaque. `encoding-type` and `fetch-owner` are accepted but ignored, and ListObjects (v1) is not implemented. GET and HEAD honor `Range` in the `bytes=first-last`, `bytes=first-`, and `bytes=-suffix` forms with a 206 status, exact `Content-Range`, and `Accept-Ranges: bytes`; unsatisfiable ranges answer 416 `InvalidRange` with `Content-Range: bytes */<length>`. PUT and CreateMultipartUpload accept `Content-Type`, `Content-Encoding`, `Content-Disposition`, `Content-Language`, `Cache-Control`, `Expires`, and `x-amz-meta-*` (names stored lowercased), and HEAD/GET replay them verbatim together with `Last-Modified` and the ETag (hex MD5 for single PUTs, `<md5-of-part-md5s>-<parts>` for multipart objects). `put_seed_object` stores an object directly, bypassing the wire, the fault script, and the journal, so a source can be seeded without polluting the assertions a scenario later makes.
+
+Fault actions cover HTTP 401/403/503 responses (`Status`), any 4xx/5xx status paired with the matching S3 error code (`ResponseStatus`), pre-dispatch delay, holding a fully computed successful response before its first byte (`Stall`), connection abort when a logical request-body threshold is reached, GetObject bodies cut off after N bytes while `Content-Length` announces the full size (`TruncateBodyAt`), streaming slow drain, and a deliberately wrong response ETag (including multipart-complete XML). `requests()` returns the ordered, credential-free request journal for assertions and `count_requests(operation, key)` counts entries for one exact key. Each record journals the `Range` and `User-Agent` request headers, the ListObjectsV2 `prefix` and `continuation-token` query values, and a `ProxyHeaderSnapshot` — the read-proxy anti-loop marker (`x-{rustfs,minio}-source-proxy-request`), the replication-check exemption header, and the client SSE-C header family (algorithm and key-MD5 values; for the key itself only its presence) — so proxy tests can pin the exact wire contract.
+
+The listener is loopback-only. It admits at most 64 active connections and two concurrently buffered request bodies; authenticated multipart-complete XML collection and assembly take both body permits. Keep-alive is disabled, request-header reads are bounded to 30 seconds, a parsed request is bounded to 65 seconds, and the complete connection lifetime is bounded to 100 seconds. It retains at most 256 buckets, 4,096 journal entries, 4,096 scripted faults, 4,096 object versions, 256 multipart uploads, and 10,000 multipart parts. Retained identifiers are capped at 1 KiB, user metadata at 2 KiB, and content type and each standard object header at 1 KiB. By default a PUT or uploaded part is capped at 64 MiB and a completed multipart object and all stored object/part data are capped at 128 MiB; `FakeS3Target::start_with_options(FakeS3TargetOptions { max_object_bytes })` raises the object cap up to 256 MiB, and the total budget then becomes twice the object cap (never below 128 MiB). Body drain, body-permit waits, delay, stall, and slow-drain execution are bounded to 30 seconds; each slow-drain slice delay must be below that bound.
@@ -15,20 +15,11 @@
 //! E2E tests for group management (fixes #2028).

 use crate::common::{RustFSTestEnvironment, admin_ok, admin_request, init_logging};
-use aws_sdk_s3::config::{Credentials, Region};
-use aws_sdk_s3::{Client, Config};
+use aws_sdk_s3::Client;
 use tracing::info;

 fn create_user_s3_client(env: &RustFSTestEnvironment, access_key: &str, secret_key: &str) -> Client {
-    let credentials = Credentials::new(access_key, secret_key, None, None, "e2e-group-test");
-    let config = Config::builder()
-        .credentials_provider(credentials)
-        .region(Region::new("us-east-1"))
-        .endpoint_url(&env.url)
-        .force_path_style(true)
-        .behavior_version_latest()
-        .build();
-    Client::from_conf(config)
+    env.create_s3_client_with_credentials(access_key, secret_key)
 }

 #[tokio::test(flavor = "multi_thread")]
@@ -15,23 +15,22 @@
 //! Four-node EC regression gate for inline storage and the inline GET reader.
 //!
 //! The storage decision is based on shard bytes (256 KiB / 32 KiB objects for
-//! the default EC 2+2 geometry), while the GET fast path has its own object-size
-//! limits (128 KiB / 16 KiB). A local OTLP/HTTP collector observes the existing
-//! reader-path counter without adding a scrape endpoint or production logging.
+//! the default EC 2+2 geometry), and the GET fast path follows the persisted
+//! inline marker. A local OTLP/HTTP collector observes the existing reader-path
+//! counter without adding a scrape endpoint or production logging.
 //! One S3 GET can select readers on multiple EC nodes, so the counter tracks
 //! distributed reader selection rather than HTTP request count.

-use crate::common::{RustFSTestClusterEnvironment, RustFSTestEnvironment, init_logging, local_http_client};
+use crate::common::{RustFSTestClusterEnvironment, RustFSTestEnvironment, init_logging};
 use aws_sdk_s3::Client;
 use aws_sdk_s3::primitives::ByteStream;
 use aws_sdk_s3::types::{
    BucketLifecycleConfiguration, BucketVersioningStatus, CompletedMultipartUpload, CompletedPart, ExpirationStatus,
    LifecycleRule, LifecycleRuleFilter, ServerSideEncryption, Transition, TransitionStorageClass, VersioningConfiguration,
 };
-use base64::Engine;
 use bytes::Bytes;
 use flate2::read::GzDecoder;
-use http::header::{CONTENT_ENCODING, HOST};
+use http::header::CONTENT_ENCODING;
 use http::{Method, Request, Response, StatusCode};
 use http_body_util::{BodyExt, Full};
 use hyper::body::Incoming;
@@ -43,9 +42,6 @@ use opentelemetry_proto::tonic::metrics::v1::{
    Metric, NumberDataPoint, ResourceMetrics, ScopeMetrics, Sum, metric, number_data_point,
 };
 use prost::Message;
-use rustfs_signer::constants::UNSIGNED_PAYLOAD;
-use rustfs_signer::sign_v4;
-use s3s::Body;
 use std::collections::BTreeMap;
 use std::convert::Infallible;
 use std::error::Error;
@@ -93,6 +89,7 @@ const MPU_PART_1_SIZE: usize = 5 * 1024 * 1024;
 const MPU_PART_2_SIZE: usize = 16 * KIB;
 const TIER_BUCKET: &str = "inline-fallback-cold-tier";
 const TIER_PREFIX: &str = "tiered";
+const ALLOW_LOOPBACK_TIER_ENDPOINT_ENV: &str = "RUSTFS_TIER_RUSTFS_ALLOW_LOOPBACK_ENDPOINT";
 const MSGPACK_FALLBACK_CONTROL_SERIES: [(&str, &str); 4] = [
    (FALLBACK_REQUEST_DIRECTION, "ReadMultipleReq"),
    (FALLBACK_RESPONSE_DIRECTION, "ReadMultipleResp"),
@@ -795,12 +792,12 @@ fn metric_attribute(key: &str, value: &str) -> KeyValue {
 }

 fn boundary_cases(state: VersionState) -> Vec<BoundaryCase> {
-    let (fast_limit, storage_limit) = match state {
-        VersionState::Enabled => (16 * KIB, 32 * KIB),
-        VersionState::Unversioned => (128 * KIB, 256 * KIB),
+    let storage_limit = match state {
+        VersionState::Enabled => 32 * KIB,
+        VersionState::Unversioned => 256 * KIB,
        // A suspended bucket stores its null version using the unversioned
        // shard threshold, while ObjectInfo keeps version-aware GET semantics.
-        VersionState::Suspended => (16 * KIB, 256 * KIB),
+        VersionState::Suspended => 256 * KIB,
    };
    let mut sizes = vec![0, 16 * KIB - 1, 16 * KIB, 16 * KIB + 1, 32 * KIB - 1, 32 * KIB, 32 * KIB + 1];
    if !matches!(state, VersionState::Enabled) {
@@ -821,7 +818,7 @@ fn boundary_cases(state: VersionState) -> Vec<BoundaryCase> {
            stored_inline: size <= storage_limit,
            expected_reader_path: if size == 0 {
                EMPTY
-            } else if size <= fast_limit {
+            } else if size <= storage_limit {
                INLINE_DIRECT
            } else {
                LEGACY_DUPLEX
@@ -1263,6 +1260,8 @@ async fn put_two_part_multipart(client: &Client, bucket: &str, key: &str) -> Tes
    Ok((body, part2, complete.e_tag().map(str::to_owned)))
 }

+/// Thin wrapper over [`crate::common::admin_request`], kept local so the call
+/// sites below keep their `Option<&str>` body shape.
 async fn signed_admin_request(
    base_url: &str,
    method: Method,
@@ -1271,30 +1270,7 @@ async fn signed_admin_request(
    access_key: &str,
    secret_key: &str,
 ) -> TestResult<(reqwest::StatusCode, String)> {
-    let url = format!("{base_url}{path}");
-    let uri = url.parse::<http::Uri>()?;
-    let authority = uri.authority().ok_or("request URL missing authority")?.to_string();
-    let body_bytes = body.map(|value| value.as_bytes().to_vec()).unwrap_or_default();
-
-    let request = http::Request::builder()
-        .method(method.clone())
-        .uri(uri)
-        .header(HOST, authority)
-        .header("x-amz-content-sha256", UNSIGNED_PAYLOAD);
-    let signed = sign_v4(request.body(Body::empty())?, 0, access_key, secret_key, "", "us-east-1");
-
-    let client = local_http_client();
-    let mut request_builder = client.request(method, url.as_str());
-    for (name, value) in signed.headers() {
-        request_builder = request_builder.header(name, value);
-    }
-    if !body_bytes.is_empty() {
-        request_builder = request_builder.body(body_bytes);
-    }
-    let response = request_builder.send().await?;
-    let status = response.status();
-    let text = response.text().await?;
-    Ok((status, text))
+    crate::common::admin_request(base_url, method, path, body.map(str::to_string), access_key, secret_key).await
 }

 fn unique_tier_name() -> String {
@@ -1808,7 +1784,7 @@ async fn four_node_inline_fallback_controls() -> TestResult {
    let collector = OtlpMetricCollector::start().await?;
    let mut cluster = RustFSTestClusterEnvironment::new(4).await?;
    configure_reader_metric_cluster(&mut cluster, &collector);
-    let sse_master_key = base64::engine::general_purpose::STANDARD.encode([0x42u8; 32]);
+    let sse_master_key = base64_simd::STANDARD.encode_to_string([0x42u8; 32]);
    cluster.set_env("RUSTFS_SSE_S3_MASTER_KEY", &sse_master_key);
    cluster.start().await?;

@@ -2017,7 +1993,7 @@ async fn four_node_mixed_msgpack_compat_mode_preserves_fallback_controls() -> Te

    let collector = OtlpMetricCollector::start().await?;
    let mut cluster = RustFSTestClusterEnvironment::new(4).await?;
-    let sse_master_key = base64::engine::general_purpose::STANDARD.encode([0x42u8; 32]);
+    let sse_master_key = base64_simd::STANDARD.encode_to_string([0x42u8; 32]);
    cluster.set_env("RUSTFS_SSE_S3_MASTER_KEY", sse_master_key);
    cluster.set_env("RUSTFS_COMPRESSION_ENABLED", "true");
    cluster.set_env("RUSTFS_COMPRESSION_MULTIPART_ENABLED", "true");
@@ -2125,6 +2101,7 @@ async fn four_node_add_tier_converges() -> TestResult {
    cold.create_s3_client().create_bucket().bucket(TIER_BUCKET).send().await?;

    let mut hot = RustFSTestClusterEnvironment::new(4).await?;
+    hot.set_env(ALLOW_LOOPBACK_TIER_ENDPOINT_ENV, "true");
    hot.start().await?;

    let tier_name = unique_tier_name();
@@ -2143,6 +2120,7 @@ async fn four_node_add_tier_converges_after_offline_node_restart_without_second_
    cold.create_s3_client().create_bucket().bucket(TIER_BUCKET).send().await?;

    let mut hot = RustFSTestClusterEnvironment::new(4).await?;
+    hot.set_env(ALLOW_LOOPBACK_TIER_ENDPOINT_ENV, "true");
    hot.start().await?;

    let tier_name = unique_tier_name();
@@ -2239,6 +2217,7 @@ async fn four_node_manual_transition_distributed_admission_conflict_reports_stat
    cold_client.create_bucket().bucket(TIER_BUCKET).send().await?;

    let mut hot = RustFSTestClusterEnvironment::new(4).await?;
+    hot.set_env(ALLOW_LOOPBACK_TIER_ENDPOINT_ENV, "true");
    hot.set_env("RUSTFS_SCANNER_ENABLED", "false");
    hot.set_env("RUSTFS_SCANNER_CYCLE", "3600");
    hot.set_env("RUSTFS_MAX_TRANSITION_WORKERS", "1");
@@ -2381,6 +2360,7 @@ async fn four_node_manual_transition_rollout_non_empty_restart_readback() -> Tes
    cold_client.create_bucket().bucket(TIER_BUCKET).send().await?;

    let mut hot = RustFSTestClusterEnvironment::new(4).await?;
+    hot.set_env(ALLOW_LOOPBACK_TIER_ENDPOINT_ENV, "true");
    hot.set_env("RUSTFS_SCANNER_ENABLED", "false");
    hot.set_env("RUSTFS_SCANNER_CYCLE", "3600");
    hot.set_env("RUSTFS_MAX_TRANSITION_WORKERS", "2");
@@ -2485,11 +2465,12 @@ async fn four_node_mixed_msgpack_compat_mode_preserves_fallback_controls_during_

    let collector = OtlpMetricCollector::start().await?;
    let mut hot = RustFSTestClusterEnvironment::new(4).await?;
+    hot.set_env(ALLOW_LOOPBACK_TIER_ENDPOINT_ENV, "true");
    configure_mixed_msgpack_cluster(&mut hot, &collector)?;
    hot.set_env("RUSTFS_SCANNER_CYCLE", "1");
    hot.set_env("RUSTFS_ILM_PROCESS_TIME", "1");

-    let sse_master_key = base64::engine::general_purpose::STANDARD.encode([0x42u8; 32]);
+    let sse_master_key = base64_simd::STANDARD.encode_to_string([0x42u8; 32]);
    hot.set_env("RUSTFS_SSE_S3_MASTER_KEY", sse_master_key);
    hot.set_env("RUSTFS_COMPRESSION_ENABLED", "true");
    hot.start().await?;
@@ -2596,6 +2577,7 @@ async fn four_node_transitioned_inline_fallback() -> TestResult {

    let collector = OtlpMetricCollector::start().await?;
    let mut hot = RustFSTestClusterEnvironment::new(4).await?;
+    hot.set_env(ALLOW_LOOPBACK_TIER_ENDPOINT_ENV, "true");
    configure_reader_metric_cluster(&mut hot, &collector);
    hot.set_env("RUSTFS_SCANNER_CYCLE", "1");
    hot.set_env("RUSTFS_ILM_PROCESS_TIME", "1");
@@ -24,9 +24,10 @@

 use crate::common::{RustFSTestEnvironment, awscurl_get, awscurl_post, init_logging as common_init_logging, local_http_client};
 use aws_sdk_s3::Client;
+use aws_sdk_s3::error::{ProvideErrorMetadata, SdkError};
 use aws_sdk_s3::primitives::ByteStream;
 use aws_sdk_s3::types::ServerSideEncryption;
-use base64::{Engine, engine::general_purpose::STANDARD as BASE64};
+use base64_simd::STANDARD as BASE64;
 use http::header::{CONTENT_TYPE, HOST};
 use md5::{Digest as Md5Digest, Md5};
 use rustfs_signer::constants::UNSIGNED_PAYLOAD;
@@ -50,6 +51,9 @@ pub const VAULT_TOKEN: &str = "dev-root-token";
 pub const VAULT_TRANSIT_PATH: &str = "transit";
 pub const VAULT_KEY_NAME: &str = "rustfs-master-key";
 pub const ENV_TEST_VAULT_BIN: &str = "RUSTFS_TEST_VAULT_BIN";
+pub const SSE_C_KEY_MISMATCH_MESSAGE: &str =
+    "The provided encryption parameters did not match the ones used originally to encrypt the object.";
+pub const SSE_C_MISSING_PARAMETERS_MESSAGE: &str = "The object was stored using a form of Server Side Encryption. The correct parameters must be provided to retrieve the object.";

 /// Initialize tracing for KMS tests with KMS-specific log levels
 pub fn init_logging() {
@@ -60,7 +64,25 @@ pub fn init_logging() {
 pub fn sse_customer_key_md5_base64(key: &str) -> String {
    let mut hasher = Md5::new();
    hasher.update(key.as_bytes());
-    BASE64.encode(hasher.finalize())
+    BASE64.encode_to_string(hasher.finalize())
+}
+
+pub fn assert_s3_error<T, E>(result: Result<T, SdkError<E>>, status: u16, code: &str, message: &str, context: &str)
+where
+    T: std::fmt::Debug,
+    E: ProvideErrorMetadata + std::fmt::Debug,
+{
+    let error = result.expect_err(context);
+    assert_eq!(
+        error.raw_response().map(|response| response.status().as_u16()),
+        Some(status),
+        "{context}: unexpected HTTP status: {error:?}"
+    );
+    let service_error = error
+        .as_service_error()
+        .expect("request failure should retain an S3 service error");
+    assert_eq!(service_error.code(), Some(code), "{context}: unexpected error code: {error:?}");
+    assert_eq!(service_error.message(), Some(message), "{context}: unexpected error message: {error:?}");
 }

 pub async fn kms_admin_request(
@@ -343,7 +365,7 @@ pub async fn create_key_with_specific_id(key_dir: &str, key_id: &str) -> Result<
        "created_at": format!("{}[UTC]", chrono::Utc::now().to_rfc3339()),
        "rotated_at": serde_json::Value::Null,
        "created_by": "e2e-test",
-        "encrypted_key_material": BASE64.encode(key_data),
+        "encrypted_key_material": BASE64.encode_to_string(key_data),
        "nonce": Vec::<u8>::new()
    });

@@ -361,7 +383,7 @@ pub async fn test_sse_c_encryption(s3_client: &Client, bucket: &str) -> Result<(
    info!("Testing SSE-C encryption");

    let test_key = "01234567890123456789012345678901"; // 32-byte key
-    let test_key_b64 = base64::engine::general_purpose::STANDARD.encode(test_key);
+    let test_key_b64 = base64_simd::STANDARD.encode_to_string(test_key);
    let test_key_md5 = sse_customer_key_md5_base64(test_key);
    let test_data = b"Hello, KMS SSE-C World!";
    let object_key = "test-sse-c-object";
@@ -529,8 +551,8 @@ pub async fn test_error_scenarios(s3_client: &Client, bucket: &str) -> Result<()
    // Test SSE-C with wrong key for download
    let test_key = "01234567890123456789012345678901";
    let wrong_key = "98765432109876543210987654321098";
-    let test_key_b64 = base64::engine::general_purpose::STANDARD.encode(test_key);
-    let wrong_key_b64 = base64::engine::general_purpose::STANDARD.encode(wrong_key);
+    let test_key_b64 = base64_simd::STANDARD.encode_to_string(test_key);
+    let wrong_key_b64 = base64_simd::STANDARD.encode_to_string(wrong_key);
    let test_key_md5 = sse_customer_key_md5_base64(test_key);
    let wrong_key_md5 = sse_customer_key_md5_base64(wrong_key);
    let test_data = b"Test data for error scenarios";
@@ -559,7 +581,13 @@ pub async fn test_error_scenarios(s3_client: &Client, bucket: &str) -> Result<()
        .send()
        .await;

-    assert!(wrong_key_result.is_err(), "Download with wrong SSE-C key should fail");
+    assert_s3_error(
+        wrong_key_result,
+        400,
+        "InvalidRequest",
+        SSE_C_KEY_MISMATCH_MESSAGE,
+        "download with a wrong SSE-C key must be rejected",
+    );
    info!("✅ Correctly rejected download with wrong SSE-C key");

    info!("Error scenario tests completed successfully");
@@ -779,7 +807,7 @@ pub async fn test_multipart_upload_with_config(
    // Prepare encryption parameters
    let (sse_c_key_b64, sse_c_key_md5) = match &config.encryption_type {
        EncryptionType::SSEC { key, key_md5 } => {
-            let key_b64 = base64::engine::general_purpose::STANDARD.encode(key);
+            let key_b64 = base64_simd::STANDARD.encode_to_string(key);
            (Some(key_b64), Some(key_md5.clone()))
        }
        _ => (None, None),
@@ -17,6 +17,7 @@

 use super::common::{
    LocalKMSTestEnvironment, VAULT_KEY_NAME, VaultTestEnvironment, configure_kms, get_kms_status, kms_admin_request, start_kms,
+    test_sse_kms_encryption,
 };
 use aws_sdk_s3::primitives::ByteStream;
 use aws_sdk_s3::types::{BucketVersioningStatus, ServerSideEncryption, VersioningConfiguration};
@@ -432,7 +433,38 @@ async fn test_configured_local_kms_admin_and_versioned_cleanup() -> TestResult {
 }

 #[tokio::test]
-#[ignore = "requires a Vault binary"]
+async fn test_admin_configured_local_kms_is_restored_after_restart() -> TestResult {
+    let mut env = LocalKMSTestEnvironment::new().await?;
+    env.base_env.start_rustfs_server(Vec::new()).await?;
+
+    let default_key_id = env.configure_local_kms().await?;
+    start_kms(&env.base_env.url, &env.base_env.access_key, &env.base_env.secret_key).await?;
+
+    env.base_env.restart_server_preserving_data(Vec::new(), &[]).await?;
+    assert_configured_status(
+        &env.base_env.url,
+        &env.base_env.access_key,
+        &env.base_env.secret_key,
+        "local",
+        &default_key_id,
+    )
+    .await?;
+
+    let bucket = format!("kms-restart-{}", Uuid::new_v4());
+    env.base_env.create_test_bucket(&bucket).await?;
+    let client = env.base_env.create_s3_client();
+    test_sse_kms_encryption(&client, &bucket).await?;
+    client
+        .delete_object()
+        .bucket(&bucket)
+        .key("test-sse-kms-object")
+        .send()
+        .await?;
+    env.base_env.delete_test_bucket(&bucket).await?;
+    Ok(())
+}
+
+#[tokio::test]
 async fn test_configured_vault_kms_admin_and_versioned_cleanup() -> TestResult {
    let mut env = VaultTestEnvironment::new().await?;
    env.start_vault().await?;
@@ -0,0 +1,140 @@
+// Copyright 2024 RustFS Team
+//
+// Licensed under the Apache License, Version 2.0 (the "License");
+// you may not use this file except in compliance with the License.
+// You may obtain a copy of the License at
+//
+//     http://www.apache.org/licenses/LICENSE-2.0
+//
+// Unless required by applicable law or agreed to in writing, software
+// distributed under the License is distributed on an "AS IS" BASIS,
+// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+// See the License for the specific language governing permissions and
+// limitations under the License.
+
+//! Ranged GETs over encrypted single-part objects.
+//!
+//! Byte-exactness must hold on every frame layout the server can write:
+//! legacy v1 (variable frames, conservative full read) and, when
+//! `RUSTFS_ENCRYPTION_FRAME_V2=true` reaches the server under test, the
+//! fixed-frame v2 layout whose marker enables the closed-form frame seek.
+//! The matrix crosses frame boundaries, starts mid-frame, and ends inside
+//! the final short frame, so a mispositioned seek cannot pass.
+
+use super::common::LocalKMSTestEnvironment;
+use crate::common::{TEST_BUCKET, init_logging};
+use aws_sdk_s3::primitives::ByteStream;
+use aws_sdk_s3::types::ServerSideEncryption;
+use tracing::info;
+
+const FRAME_PLAINTEXT: usize = 8 * 1024;
+
+#[tokio::test]
+async fn sse_s3_single_part_ranged_gets_are_byte_exact() -> Result<(), Box<dyn std::error::Error + Send + Sync>> {
+    init_logging();
+    info!("Testing ranged GETs over an SSE-S3 single-part object");
+
+    let mut kms_env = LocalKMSTestEnvironment::new().await?;
+    let _default_key_id = kms_env.start_rustfs_for_local_kms().await?;
+    kms_env.wait_for_kms_ready().await?;
+
+    let s3_client = kms_env.base_env.create_s3_client();
+    kms_env.base_env.create_test_bucket(TEST_BUCKET).await?;
+
+    let test_key = "encrypted-range-get";
+    let body: Vec<u8> = (0..3 * FRAME_PLAINTEXT + 500).map(|i| (i % 251) as u8).collect();
+
+    let put = s3_client
+        .put_object()
+        .bucket(TEST_BUCKET)
+        .key(test_key)
+        .server_side_encryption(ServerSideEncryption::Aes256)
+        .body(ByteStream::from(body.clone()))
+        .send()
+        .await?;
+    assert_eq!(
+        put.server_side_encryption(),
+        Some(&ServerSideEncryption::Aes256),
+        "the object under test must actually be encrypted"
+    );
+
+    let cases: &[(usize, usize)] = &[
+        // Head range inside frame 0.
+        (0, 99),
+        // Crossing the first frame boundary.
+        (FRAME_PLAINTEXT - 1, FRAME_PLAINTEXT),
+        // Starting exactly on a frame boundary.
+        (FRAME_PLAINTEXT, FRAME_PLAINTEXT + 9),
+        // Mid-object, mid-frame on both ends.
+        (2 * FRAME_PLAINTEXT + 5, 3 * FRAME_PLAINTEXT + 100),
+        // Tail range ending inside the final short frame.
+        (3 * FRAME_PLAINTEXT + 100, 3 * FRAME_PLAINTEXT + 499),
+    ];
+
+    for &(start, end) in cases {
+        let response = s3_client
+            .get_object()
+            .bucket(TEST_BUCKET)
+            .key(test_key)
+            .range(format!("bytes={start}-{end}"))
+            .send()
+            .await?;
+        assert_eq!(
+            response.content_length(),
+            Some((end - start + 1) as i64),
+            "range {start}-{end} content length"
+        );
+        let data = response.body.collect().await?.into_bytes();
+        assert_eq!(data.as_ref(), &body[start..=end], "range {start}-{end} must be byte-exact");
+    }
+
+    // A suffix range exercises the offset resolution path as well.
+    let response = s3_client
+        .get_object()
+        .bucket(TEST_BUCKET)
+        .key(test_key)
+        .range("bytes=-123")
+        .send()
+        .await?;
+    let data = response.body.collect().await?.into_bytes();
+    assert_eq!(data.as_ref(), &body[body.len() - 123..], "suffix range must be byte-exact");
+
+    // The unranged body still round-trips.
+    let response = s3_client.get_object().bucket(TEST_BUCKET).key(test_key).send().await?;
+    let data = response.body.collect().await?.into_bytes();
+    assert_eq!(data.as_ref(), body.as_slice(), "full body must round-trip");
+
+    // A block-aligned object ends in an empty authenticated final frame under
+    // the v2 layout; tail ranges touching the last plaintext byte must not be
+    // misread as truncation.
+    let aligned_key = "encrypted-range-get-aligned";
+    let aligned_body: Vec<u8> = (0..3 * FRAME_PLAINTEXT).map(|i| ((i + 3) % 251) as u8).collect();
+    s3_client
+        .put_object()
+        .bucket(TEST_BUCKET)
+        .key(aligned_key)
+        .server_side_encryption(ServerSideEncryption::Aes256)
+        .body(ByteStream::from(aligned_body.clone()))
+        .send()
+        .await?;
+    for (start, end) in [
+        (2 * FRAME_PLAINTEXT + 10, 3 * FRAME_PLAINTEXT - 1),
+        (3 * FRAME_PLAINTEXT - 1, 3 * FRAME_PLAINTEXT - 1),
+    ] {
+        let response = s3_client
+            .get_object()
+            .bucket(TEST_BUCKET)
+            .key(aligned_key)
+            .range(format!("bytes={start}-{end}"))
+            .send()
+            .await?;
+        let data = response.body.collect().await?.into_bytes();
+        assert_eq!(
+            data.as_ref(),
+            &aligned_body[start..=end],
+            "aligned range {start}-{end} must be byte-exact"
+        );
+    }
+
+    Ok(())
+}
@@ -0,0 +1,220 @@
+// Copyright 2024 RustFS Team
+//
+// Licensed under the Apache License, Version 2.0 (the "License");
+// you may not use this file except in compliance with the License.
+// You may obtain a copy of the License at
+//
+//     http://www.apache.org/licenses/LICENSE-2.0
+//
+// Unless required by applicable law or agreed to in writing, software
+// distributed under the License is distributed on an "AS IS" BASIS,
+// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+// See the License for the specific language governing permissions and
+// limitations under the License.
+
+//! Anonymous access to SSE-KMS objects under per-key authorization.
+//!
+//! Locks both halves of the anonymous contract decided in backlog#2028 (D4):
+//!
+//! - **Enforcement on**: anonymous requests hold no `kms` grants, so a public
+//!   bucket policy does not let them read SSE-KMS objects or write through an
+//!   SSE-KMS default-encryption rule. Both fail with `AccessDenied`.
+//! - **Enforcement off** (the default): bucket policy alone governs anonymous
+//!   access, matching the pre-enforcement behavior — public SSE-KMS objects are
+//!   decrypted and served, and anonymous writes are encrypted under the default
+//!   key.
+//!
+//! The denial today is emergent — an empty-account principal falling through to
+//! the IAM default deny — so without this file a refactor of principal
+//! construction or policy evaluation could silently flip it. Each test carries a
+//! plaintext-object positive control: a denial proves nothing while the bucket
+//! policy has not propagated.
+
+use super::common::{LocalKMSTestEnvironment, create_key_with_specific_id};
+use crate::common::{init_logging, local_http_client};
+use aws_sdk_s3::primitives::ByteStream;
+use aws_sdk_s3::types::{
+    ServerSideEncryption, ServerSideEncryptionByDefault, ServerSideEncryptionConfiguration, ServerSideEncryptionRule,
+};
+use std::time::Duration;
+
+type TestResult = Result<(), Box<dyn std::error::Error + Send + Sync>>;
+
+const DEFAULT_KEY: &str = "kms-anon-default-key";
+const BUCKET: &str = "kms-anon-enforcement";
+const PLAIN_OBJECT: &str = "plain.txt";
+const ENCRYPTED_OBJECT: &str = "encrypted.txt";
+const PAYLOAD: &[u8] = b"kms anonymous enforcement payload";
+
+/// How long a bucket policy change may take to reach the request path.
+const POLICY_PROPAGATION: Duration = Duration::from_secs(20);
+
+/// Start a local-KMS server and build the public-bucket fixture.
+///
+/// The bucket holds a plaintext object (the positive control), an SSE-KMS
+/// object, an SSE-KMS default-encryption rule, and a bucket policy opening
+/// `GetObject`/`PutObject` to everyone. The enforcement switch defaults to off,
+/// so the enforcing case has to set it explicitly.
+async fn start_public_sse_kms_bucket(env: &mut LocalKMSTestEnvironment, enforce: bool) -> TestResult {
+    create_key_with_specific_id(&env.kms_keys_dir, DEFAULT_KEY).await?;
+
+    let key_dir = env.kms_keys_dir.clone();
+    let args = vec![
+        "--kms-enable",
+        "--kms-backend",
+        "local",
+        "--kms-key-dir",
+        key_dir.as_str(),
+        "--kms-default-key-id",
+        DEFAULT_KEY,
+    ];
+    let mut envs = vec![("RUSTFS_KMS_ALLOW_INSECURE_DEV_DEFAULTS", "true")];
+    if enforce {
+        envs.push(("RUSTFS_KMS_ENFORCE_SSE_KEY_POLICY", "true"));
+    }
+    env.base_env.start_rustfs_server_with_env(args, &envs).await?;
+    env.base_env.create_test_bucket(BUCKET).await?;
+
+    let owner = env.base_env.create_s3_client();
+
+    owner
+        .put_object()
+        .bucket(BUCKET)
+        .key(PLAIN_OBJECT)
+        .body(ByteStream::from_static(PAYLOAD))
+        .send()
+        .await?;
+    owner
+        .put_object()
+        .bucket(BUCKET)
+        .key(ENCRYPTED_OBJECT)
+        .body(ByteStream::from_static(PAYLOAD))
+        .server_side_encryption(ServerSideEncryption::AwsKms)
+        .ssekms_key_id(DEFAULT_KEY)
+        .send()
+        .await?;
+
+    let encryption_config = ServerSideEncryptionConfiguration::builder()
+        .rules(
+            ServerSideEncryptionRule::builder()
+                .apply_server_side_encryption_by_default(
+                    ServerSideEncryptionByDefault::builder()
+                        .sse_algorithm(ServerSideEncryption::AwsKms)
+                        .kms_master_key_id(DEFAULT_KEY)
+                        .build()?,
+                )
+                .build(),
+        )
+        .build()?;
+    owner
+        .put_bucket_encryption()
+        .bucket(BUCKET)
+        .server_side_encryption_configuration(encryption_config)
+        .send()
+        .await?;
+
+    let policy = serde_json::json!({
+        "Version": "2012-10-17",
+        "Statement": [{
+            "Sid": "PublicReadWrite",
+            "Effect": "Allow",
+            "Principal": "*",
+            "Action": ["s3:GetObject", "s3:PutObject"],
+            "Resource": [format!("arn:aws:s3:::{BUCKET}/*")]
+        }]
+    })
+    .to_string();
+    owner.put_bucket_policy().bucket(BUCKET).policy(&policy).send().await?;
+    let _ = owner.delete_public_access_block().bucket(BUCKET).send().await;
+
+    Ok(())
+}
+
+fn object_url(env: &LocalKMSTestEnvironment, key: &str) -> String {
+    format!("{}/{BUCKET}/{key}", env.base_env.url)
+}
+
+async fn anonymous_get(env: &LocalKMSTestEnvironment, key: &str) -> Result<reqwest::Response, reqwest::Error> {
+    local_http_client().get(object_url(env, key)).send().await
+}
+
+async fn anonymous_put(env: &LocalKMSTestEnvironment, key: &str) -> Result<reqwest::Response, reqwest::Error> {
+    local_http_client().put(object_url(env, key)).body(PAYLOAD).send().await
+}
+
+/// Retry the plaintext read until the public bucket policy is live.
+async fn wait_for_public_read(env: &LocalKMSTestEnvironment) -> TestResult {
+    let deadline = tokio::time::Instant::now() + POLICY_PROPAGATION;
+    loop {
+        let status = anonymous_get(env, PLAIN_OBJECT).await?.status();
+        if status.as_u16() == 200 {
+            return Ok(());
+        }
+        if tokio::time::Instant::now() >= deadline {
+            return Err(format!("positive control never became readable: anonymous GET {PLAIN_OBJECT} -> {status}").into());
+        }
+        tokio::time::sleep(Duration::from_millis(500)).await;
+    }
+}
+
+async fn assert_anonymous_denied(response: reqwest::Response, what: &str) -> TestResult {
+    let status = response.status().as_u16();
+    let body = response.text().await?;
+    assert_eq!(status, 403, "{what} must be denied, got {status}: {body}");
+    assert!(body.contains("AccessDenied"), "{what} must carry AccessDenied: {body}");
+    Ok(())
+}
+
+/// Enforcement on: a public bucket policy does not exempt anonymous requests
+/// from per-key authorization, on either the read or the default-encryption
+/// write path.
+#[tokio::test(flavor = "multi_thread")]
+async fn anonymous_sse_kms_denied_under_enforcement() -> TestResult {
+    init_logging();
+
+    let mut env = LocalKMSTestEnvironment::new().await?;
+    start_public_sse_kms_bucket(&mut env, true).await?;
+    wait_for_public_read(&env).await?;
+
+    let read = anonymous_get(&env, ENCRYPTED_OBJECT).await?;
+    assert_anonymous_denied(read, "anonymous GET of an SSE-KMS object").await?;
+
+    let write = anonymous_put(&env, "anon-write.txt").await?;
+    assert_anonymous_denied(write, "anonymous PUT through an SSE-KMS default-encryption rule").await?;
+
+    Ok(())
+}
+
+/// Enforcement off (the default): bucket policy alone governs anonymous access,
+/// and the default-encryption rule still encrypts anonymous writes.
+#[tokio::test(flavor = "multi_thread")]
+async fn anonymous_sse_kms_governed_by_bucket_policy_without_enforcement() -> TestResult {
+    init_logging();
+
+    let mut env = LocalKMSTestEnvironment::new().await?;
+    start_public_sse_kms_bucket(&mut env, false).await?;
+    wait_for_public_read(&env).await?;
+
+    let read = anonymous_get(&env, ENCRYPTED_OBJECT).await?;
+    assert_eq!(read.status().as_u16(), 200, "anonymous GET of a public SSE-KMS object must succeed");
+    assert_eq!(read.bytes().await?.as_ref(), PAYLOAD, "the object must be served decrypted");
+
+    let write = anonymous_put(&env, "anon-write.txt").await?;
+    assert_eq!(write.status().as_u16(), 200, "anonymous PUT to a public bucket must succeed");
+
+    let stored = env
+        .base_env
+        .create_s3_client()
+        .head_object()
+        .bucket(BUCKET)
+        .key("anon-write.txt")
+        .send()
+        .await?;
+    assert_eq!(
+        stored.server_side_encryption(),
+        Some(&ServerSideEncryption::AwsKms),
+        "the anonymous write must be encrypted by the bucket default rule"
+    );
+
+    Ok(())
+}
@@ -19,9 +19,9 @@
 //! complex workflows.

 use super::common::{
-    EncryptionType, LocalKMSTestEnvironment, MultipartTestConfig, create_sse_c_config, sse_customer_key_md5_base64,
-    test_all_multipart_encryption_types, test_kms_key_management, test_multipart_upload_with_config, test_sse_c_encryption,
-    test_sse_kms_encryption, test_sse_s3_encryption,
+    EncryptionType, LocalKMSTestEnvironment, MultipartTestConfig, SSE_C_KEY_MISMATCH_MESSAGE, assert_s3_error,
+    create_sse_c_config, sse_customer_key_md5_base64, test_all_multipart_encryption_types, test_kms_key_management,
+    test_multipart_upload_with_config, test_sse_c_encryption, test_sse_kms_encryption, test_sse_s3_encryption,
 };
 use crate::common::{TEST_BUCKET, init_logging};
 use tracing::info;
@@ -177,7 +177,7 @@ async fn test_comprehensive_key_isolation() -> Result<(), Box<dyn std::error::Er
    // Verify that files cannot be read with wrong keys
    info!("🔒 Verify key isolation");
    let wrong_key = "11111111111111111111111111111111";
-    let wrong_key_b64 = base64::Engine::encode(&base64::engine::general_purpose::STANDARD, wrong_key);
+    let wrong_key_b64 = base64_simd::STANDARD.encode_to_string(wrong_key);
    let wrong_key_md5 = sse_customer_key_md5_base64(wrong_key);

    // Try to read file encrypted with key1 using wrong key
@@ -191,7 +191,13 @@ async fn test_comprehensive_key_isolation() -> Result<(), Box<dyn std::error::Er
        .send()
        .await;

-    assert!(wrong_read_result.is_err(), "The encrypted file should not be readable with the wrong key");
+    assert_s3_error(
+        wrong_read_result,
+        400,
+        "InvalidRequest",
+        SSE_C_KEY_MISMATCH_MESSAGE,
+        "multipart SSE-C object GET with a wrong key must be rejected",
+    );
    info!("✅ Confirm that key isolation is working correctly");

    kms_env.base_env.delete_test_bucket(TEST_BUCKET).await?;
@@ -21,21 +21,13 @@
 //! - Concurrent encryption operations
 //! - Security validation tests

-use super::common::{LocalKMSTestEnvironment, sse_customer_key_md5_base64};
+use super::common::{LocalKMSTestEnvironment, SSE_C_KEY_MISMATCH_MESSAGE, assert_s3_error, sse_customer_key_md5_base64};
 use crate::common::{TEST_BUCKET, init_logging};
 use aws_sdk_s3::types::ServerSideEncryption;
-use base64::Engine;
-use md5::{Digest as Md5Digest, Md5};
 use std::sync::Arc;
 use tokio::sync::Semaphore;
 use tracing::{info, warn};

-fn md5_hex(input: impl AsRef<[u8]>) -> String {
-    let mut hasher = Md5::new();
-    hasher.update(input.as_ref());
-    hex::encode(hasher.finalize())
-}
-
 /// Test encryption of zero-byte files (empty files)
 #[tokio::test]
 async fn test_kms_zero_byte_file_encryption() -> Result<(), Box<dyn std::error::Error + Send + Sync>> {
@@ -75,7 +67,7 @@ async fn test_kms_zero_byte_file_encryption() -> Result<(), Box<dyn std::error::
    // Test SSE-C with zero-byte file
    info!("📤 Testing SSE-C with zero-byte file");
    let test_key = "01234567890123456789012345678901";
-    let test_key_b64 = base64::engine::general_purpose::STANDARD.encode(test_key);
+    let test_key_b64 = base64_simd::STANDARD.encode_to_string(test_key);
    let test_key_md5 = sse_customer_key_md5_base64(test_key);
    let object_key_c = "zero-byte-sse-c";

@@ -168,7 +160,7 @@ async fn test_kms_single_byte_file_encryption() -> Result<(), Box<dyn std::error
    // Test SSE-C with single byte
    info!("📤 Testing SSE-C with single-byte file");
    let test_key = "01234567890123456789012345678901";
-    let test_key_b64 = base64::engine::general_purpose::STANDARD.encode(test_key);
+    let test_key_b64 = base64_simd::STANDARD.encode_to_string(test_key);
    let test_key_md5 = sse_customer_key_md5_base64(test_key);
    let object_key_c = "single-byte-sse-c";

@@ -294,8 +286,8 @@ async fn test_kms_invalid_key_scenarios() -> Result<(), Box<dyn std::error::Erro
    // Test 1: Invalid key length for SSE-C
    info!("🔍 Testing invalid SSE-C key length");
    let invalid_short_key = "short"; // Too short
-    let invalid_key_b64 = base64::engine::general_purpose::STANDARD.encode(invalid_short_key);
-    let invalid_key_md5 = md5_hex(invalid_short_key);
+    let invalid_key_b64 = base64_simd::STANDARD.encode_to_string(invalid_short_key);
+    let invalid_key_md5 = sse_customer_key_md5_base64(invalid_short_key);

    let invalid_key_result = s3_client
        .put_object()
@@ -308,14 +300,32 @@ async fn test_kms_invalid_key_scenarios() -> Result<(), Box<dyn std::error::Erro
        .send()
        .await;

-    assert!(invalid_key_result.is_err(), "Should reject invalid key length");
+    assert_s3_error(
+        invalid_key_result,
+        400,
+        "InvalidRequest",
+        "SSE-C key must be 32 bytes (256 bits), got 5 bytes.",
+        "invalid SSE-C key length must be rejected",
+    );
+    assert_s3_error(
+        s3_client
+            .get_object()
+            .bucket(TEST_BUCKET)
+            .key("test-invalid-key-length")
+            .send()
+            .await,
+        404,
+        "NoSuchKey",
+        "The specified key does not exist.",
+        "rejected invalid-key PUT must not create an object",
+    );
    info!("✅ Correctly rejected invalid key length");

    // Test 2: Mismatched MD5 for SSE-C
    info!("🔍 Testing mismatched MD5 for SSE-C key");
    let valid_key = "01234567890123456789012345678901";
-    let valid_key_b64 = base64::engine::general_purpose::STANDARD.encode(valid_key);
-    let wrong_md5 = "wrongmd5hash12345678901234567890"; // Wrong MD5
+    let valid_key_b64 = base64_simd::STANDARD.encode_to_string(valid_key);
+    let wrong_md5 = sse_customer_key_md5_base64("98765432109876543210987654321098");

    let wrong_md5_result = s3_client
        .put_object()
@@ -324,11 +334,24 @@ async fn test_kms_invalid_key_scenarios() -> Result<(), Box<dyn std::error::Erro
        .body(aws_sdk_s3::primitives::ByteStream::from(test_data.to_vec()))
        .sse_customer_algorithm("AES256")
        .sse_customer_key(&valid_key_b64)
-        .sse_customer_key_md5(wrong_md5)
+        .sse_customer_key_md5(&wrong_md5)
        .send()
        .await;

-    assert!(wrong_md5_result.is_err(), "Should reject mismatched MD5");
+    assert_s3_error(
+        wrong_md5_result,
+        400,
+        "InvalidRequest",
+        "The calculated MD5 hash of the key did not match the hash that was provided.",
+        "mismatched SSE-C key MD5 must be rejected",
+    );
+    assert_s3_error(
+        s3_client.get_object().bucket(TEST_BUCKET).key("test-wrong-md5").send().await,
+        404,
+        "NoSuchKey",
+        "The specified key does not exist.",
+        "rejected mismatched-MD5 PUT must not create an object",
+    );
    info!("✅ Correctly rejected mismatched MD5");

    // Test 3: Try to access SSE-C object without providing key
@@ -355,7 +378,28 @@ async fn test_kms_invalid_key_scenarios() -> Result<(), Box<dyn std::error::Erro
        .send()
        .await;

-    assert!(no_key_result.is_err(), "Should require SSE-C key for access");
+    assert_s3_error(
+        no_key_result,
+        400,
+        "InvalidRequest",
+        "The object was stored using a form of Server Side Encryption. The correct parameters must be provided to retrieve the object.",
+        "SSE-C object GET without a customer key must be rejected",
+    );
+
+    let recovered = s3_client
+        .get_object()
+        .bucket(TEST_BUCKET)
+        .key("test-sse-c-no-key-access")
+        .sse_customer_algorithm("AES256")
+        .sse_customer_key(&valid_key_b64)
+        .sse_customer_key_md5(&valid_key_md5)
+        .send()
+        .await?
+        .body
+        .collect()
+        .await?
+        .into_bytes();
+    assert_eq!(recovered.as_ref(), test_data, "failed GET must not corrupt the SSE-C object");
    info!("✅ Correctly required SSE-C key for access");

    kms_env.base_env.delete_test_bucket(TEST_BUCKET).await?;
@@ -420,7 +464,7 @@ async fn test_kms_concurrent_encryption() -> Result<(), Box<dyn std::error::Erro
                2 => {
                    // SSE-C
                    let key = format!("testkey{i:026}"); // 32-byte key
-                    let key_b64 = base64::engine::general_purpose::STANDARD.encode(&key);
+                    let key_b64 = base64_simd::STANDARD.encode_to_string(&key);
                    let key_md5 = sse_customer_key_md5_base64(&key);

                    client
@@ -490,8 +534,8 @@ async fn test_kms_key_validation_security() -> Result<(), Box<dyn std::error::Er
    let key1 = "key1key1key1key1key1key1key1key1"; // 32 bytes
    let key2 = "key2key2key2key2key2key2key2key2"; // 32 bytes

-    let key1_b64 = base64::engine::general_purpose::STANDARD.encode(key1);
-    let key2_b64 = base64::engine::general_purpose::STANDARD.encode(key2);
+    let key1_b64 = base64_simd::STANDARD.encode_to_string(key1);
+    let key2_b64 = base64_simd::STANDARD.encode_to_string(key2);
    let key1_md5 = sse_customer_key_md5_base64(key1);
    let key2_md5 = sse_customer_key_md5_base64(key2);

@@ -563,7 +607,13 @@ async fn test_kms_key_validation_security() -> Result<(), Box<dyn std::error::Er
        .send()
        .await;

-    assert!(wrong_key_result.is_err(), "Should not be able to decrypt with wrong key");
+    assert_s3_error(
+        wrong_key_result,
+        400,
+        "InvalidRequest",
+        SSE_C_KEY_MISMATCH_MESSAGE,
+        "SSE-C object GET with the wrong customer key must be rejected",
+    );
    info!("✅ Key isolation verified - wrong key cannot decrypt data");

    kms_env.base_env.delete_test_bucket(TEST_BUCKET).await?;
--- a/Show More
+++ b/Show More