55 lines
3.4 KiB
Markdown
55 lines
3.4 KiB
Markdown
---
|
|
id: "bug-1943"
|
|
status: "fixed"
|
|
status_updated: "2026-10-10"
|
|
status_note: ""
|
|
component: "scheduler"
|
|
severity: "wrong-result"
|
|
timestamp: "2026-10-10T08:14:54Z"
|
|
last_seen: "2026-10-10T08:14:54Z"
|
|
occurrences: 1
|
|
files: ["src/scheduler/src/core_events.c", "src/scheduler/src/scheduler.c"]
|
|
tags: ["scheduler", "sim", "equivalence", "limit"]
|
|
related: []
|
|
fix_commits: ["8b89a0ce"]
|
|
gitea_issue: 374
|
|
gitea_synced_status: "fixed"
|
|
gitea_synced_hash: "68ef4f5dc713"
|
|
---
|
|
|
|
# bug-1943: xce_scheduler --sim does not load user and host groups, so group-scoped limits differ from live
|
|
|
|
## Description
|
|
**Observed:** (source reading, `.kiro/reviews/project-gap-review-2026-10-10.md` (P0); to be confirmed by a failing test before the fix)
|
|
On a config push the live scheduler loads user and host groups (`groupmgr_load_configs`, `src/scheduler/src/scheduler.c:1144-1145`); the core's config event used by the simulator applies only resources, resource values and queues (`core_events.c:846-850`).
|
|
|
|
**Expected:** Live and simulated scheduling apply the same groups (scheduler-driver-equivalence.md).
|
|
**Impact:** Replays of clusters with group-scoped limits make different decisions than the live scheduler.
|
|
**Found by:** review: `.kiro/reviews/project-gap-review-2026-10-10.md` (P0)
|
|
## Environment
|
|
- OS: no /etc/os-release; Docker container; NixOS kernel; kernel Linux 6.18.33 x86_64
|
|
- libc: glibc 2.42-47 (interpreter /nix/store/wb6rhpznjfczwlwx23zmdrrw74bayxw4-glibc-2.42-47/lib/ld-linux-x86-64.so.2)
|
|
- Toolchain: gcc (GCC) 15.2.0, meson 1.9.1
|
|
- Build: asan (xce.build)
|
|
- Code: scheduler-framework @ 94472ee3-dirty
|
|
- Deployment: any (found by source reading; the failing test that confirms it is named in Verification)
|
|
|
|
## Reproduction
|
|
1. Record a session with a limit scoped to a user group, replay it with `xce_scheduler --sim`.
|
|
2. The limit does not match the group's users in sim.
|
|
- Frequency: deterministic (by code reading; confirmed by the RED test of the fix)
|
|
- Evidence: the code sites quoted in Description; no logs kept.
|
|
## Root cause
|
|
Only `sched_init_config()` (`scheduler.c`, live) called `groupmgr_load_configs`; the core `CONFIG_PUSH` event (`core_events.c:xce_core_push_events`) and the replay snapshot loader (`snapshot_load.c:xce_sched_core_load_snapshot`, from which `sim_bootstrap_limit_plugin` seeds limits) never loaded groups. Pre-existing.
|
|
|
|
## Fix
|
|
`core.h`/`core.c`: new `xce_core_load_groups()` is the single loader, called from `sched_init_config()` (replacing its direct calls), the `CONFIG_PUSH` event (first) and `snapshot_load.c` (before the limit plugin is seeded); `snapshot.c` field-completeness note updated. Follow-up: limits are not re-seeded on a mid-replay CONFIG_PUSH.
|
|
|
|
## Verification
|
|
- Test: `test_user_group_limit_matches_in_sim_as_in_live` (`src/scheduler/testsuite/test_sim_group_limit_parity.c`): RED `:318: FAIL: Expected 1 Was 0 (sim must match the user-group-scoped limit)`, and with only the CONFIG_PUSH part reverted `:334: FAIL: CONFIG_PUSH must register the config's user groups`, GREEN after.
|
|
- Suites: `xce-build.sh all asan xce.build --incremental` on the merged branch, 2026-10-10: all C suites Ok 155 / Fail 0, frontend Go unit tests ok, xceadmin tests ok; apiserver `CGO_ENABLED=0 go test ./...` (incl. middleware, router, cmd) ok
|
|
- Live: n/a: unit/fixture-level verification only; no live deployment was run (follow-up)
|
|
|
|
## References
|
|
- Fix commit: `8b89a0ce` (on `scheduler-framework`; worker-branch original in the merge)
|
|
- Review: `.kiro/reviews/project-gap-review-2026-10-10.md`
|