myxce/.wolf/bugs/bug-1938.md

id	status	status_updated	status_note	component	severity	timestamp	last_seen	occurrences	files	tags	related	fix_commits	gitea_issue	gitea_synced_status	gitea_synced_hash
bug-1938	fixed	2026-10-10		scheduler	wrong-result	2026-10-10T08:14:41Z	2026-10-10T08:14:41Z	1	src/scheduler/src/core.c src/scheduler/src/plugins/builtin_fcfs.c	scheduler queue admin-control		f51a19d3	369	fixed	207d5b7a1bef

bug-1938: xce_scheduler dispatches jobs from closed or inactive queues (queue readiness gate is a stub)

Description

Observed: (source reading, .kiro/reviews/project-gap-review-2026-10-10.md (P0); to be confirmed by a failing test before the fix) The admin queue controls set and clear the closed/inactive flags (src/controller/queue.c:131-152), but the scheduler's gates always pass:

/* FIXME this is just stub, we need to add more */
int sched_job_is_ready(xce_sched_job_t* j) { return 1; }
/* FIXME this is just stub, we need to add more */
int sched_queue_is_ready(xce_sched_queue_t* q) { return 1; }   /* core.c:646-649 */

Expected: Jobs of an inactive queue are not dispatched (they stay pending with a reason); a closed queue accepts no new submissions. Impact: Admin queue control silently ignored: inactivating a queue does not stop dispatch. 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. Deploy a cluster, inactivate queue normal (queue control), submit a job to it.
2. The job is dispatched and runs.

• 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

sched_queue_is_ready() (src/scheduler/src/core.c) was a return 1 stub, the only queue gate of both pickers (builtin_fcfs_next_job, fcfs_elect_job); the queue status bits did reach xce_sched_queue_t. Pre-existing stub.

Fix

core.c: sched_queue_is_ready() returns false for QUEUE_STAT_INACT_ADMIN or QUEUE_STAT_INACT_WINDOW; QUEUE_STAT_CLOSED is deliberately not a dispatch gate (LSF: a closed queue still dispatches its pending jobs; closing blocks submissions). Live and sim share the core and picker. Follow-ups: the controller does not reject submissions to a closed queue and the apiserver skips the check for the default queue; no pend reason for held jobs.

Verification

• Test: test_inactive_queue_holds_jobs_closed_queue_dispatches (src/scheduler/testsuite/test_queue_admin_control.c): RED :208: FAIL: Expected 2 Was 3 (an inactive queue dispatches nothing), GREEN after on both fixture shapes.
• 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: f51a19d3 (on scheduler-framework; worker-branch original in the merge)
• Review: .kiro/reviews/project-gap-review-2026-10-10.md