Zhengchao AnandGitHub 16a91c35ec perf(iam): load IAM lists in fixed chunks to avoid O(n^2) startup load (backlog#806) (#4537) ...

load_all fetched policies/users/user-policies by passing the FULL remaining
list to load_*_concurrent every iteration and then split_off(32), so each item
was re-fetched once per preceding chunk — O(n^2) redundant loads on startup for
>32 items. Replace the split_off loops with chunks(32) so each item is fetched
exactly once. Behavior-preserving (cache inserts were already idempotent).

2026-07-09 02:01:36 +08:00

..

object.rs

perf(iam): load IAM lists in fixed chunks to avoid O(n^2) startup load (backlog#806) (#4537)

2026-07-09 02:01:36 +08:00