Rongsong Shen
d26ff8870d
The buglog was one 27K-line JSON file whose fields covered only symptom,
root cause and fix, as free text: 2% of entries named the code version a
bug was found on, 12% the build profile, 36% a way to reproduce, and 58% of
the file was OpenWolf hook auto-detections ("Missing guard clause",
"Function not marked async" on .md files).
- .claude/skills/buglog/bug-record.md: the record standard. Eight
sections (Summary as the title, Description, Environment, Reproduction;
Root cause, Fix, Verification, References once fixed), each with the
content it must contain -- e.g. Environment needs the OS, the libc the
failing binary loads (from its ELF interpreter: glibc 2.42 in the nix
shell vs 2.40 on the host), toolchain, build profile and commit.
- .wolf/bugs/<id>.md: 500 records migrated from buglog.json; every field
verified to round-trip. They carry legacy: true and an explicit "not
recorded" where the old log had no such field; component derived from
file paths, severity from keywords (else unknown), fix_commits from
hashes in status_note/fix that git resolves (91 of 463 fixed).
- .wolf/buglog-archive.json: the 710 hook auto-detections, verbatim.
- .wolf/buglog.json is now an index generated from the records, keeping
the fields OpenWolf's hooks and `openwolf bug search` read.
- scripts/buglog.py rewritten: new (template + Environment collected from
the machine and --binary), check, status (refuses a status the record
doesn't support; fixed needs Root cause, Fix, Verification and
--commit hashes git resolves), note, recur, set, show, list, triage,
env, index. 29 unittest cases in scripts/tests/test_buglog.py.
- scripts/git-hooks/pre-commit runs `check --staged`: incomplete staged
records, a legacy record hand-changed to fixed without the fix-time
sections, a new record claiming legacy: true, or a stale index refuse
the commit.
- scripts/gitea-sync.py reads the records: issue body = the record's
sections plus linked fix commits (origin remote, GITEA_COMMIT_REPO to
override), and a body hash so an issue's text is refreshed when the
record changes.
- bug-1866 rewritten to the full standard as the worked example.
Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
3.5 KiB
| id | status | status_updated | status_note | component | severity | timestamp | last_seen | occurrences | files | tags | related | fix_commits | legacy | |||||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| bug-1624 | fixed | 2026-09-24 | scheduler | unknown | 2026-08-22T19:53:00Z | 2026-09-19T09:00:00Z | 2 |
|
|
|
|
true |
bug-1624: None yet observed at runtime -- found via code trace while answering an architectural question…
Description
None yet observed at runtime -- found via code trace while answering an architectural question about job RUN-state boundaries and over/under-commit safety.
Environment
Not recorded (migrated from buglog.json).
Reproduction
Not recorded (migrated from buglog.json).
Root cause
sched_account_jobs() (scheduler.c:1258-1292) re-derives core->sharedInUse/worker->resInUse every round from scratch, but gates on JOB_IS_RUN(job->state) before checking job->accounts. A job that has been DISPATCHED (job->accounts set synchronously at commit-decision time via sched_commit_job_alloc(), scheduler.c:1446) but not yet confirmed RUN by the worker is still job->state==PEND, so it is skipped by this resum. Since xce_core_begin_session() (core_session.c:66-108) unconditionally zeroes sharedInUse/resInUse every round (normal-mode cadence: 1s, main.c:453-460), that job's charge -- correct within the round it was dispatched -- is wiped at the very next round and not restored until the RUN confirmation is processed. The admission check other jobs go through (sched_check_host_satisfy_/sched_check_shared_satisfy_, resource.c:252-370, called from seqjob_allocator.c:130) reads exactly this same resInUse/sharedInUse, with no separate reserved/committed pool (xce_sched_job_t has only state/allocs/accounts, sched/types.h:96-116) -- so once wiped, the resource looks free again and can be granted to a second job, a genuine over-commit/double-booking risk under any dispatch-to-worker-confirmation latency exceeding ~1 round (ordinary for real worker RPC + fork/exec).
Fix
FIXED (2026-08-23, scheduler-dispatch-commit tasks 2.1 + 3.1). Scheduler side (task 2.1, commit 8607348c): sched_commit_decision()'s DISPATCH branch now sets job->state = JOB_STAT_RUN synchronously, in the same operation that already sets job->accounts/job->allocs -- no separate fix to sched_account_jobs()'s JOB_IS_RUN gate was needed; the job simply reaches RUN from the round it's dispatched, so the existing gate naturally includes it every subsequent round regardless of confirmation latency. Controller side (task 3.1, commit d9aedc4b): dispatch_a_job_() gates the job.allocs commit + JOB_STAT_RUN transition together on xce_startjob_req() succeeding, keeping both sides consistent. Regression-tested end-to-end: test_dispatch_commit_survives_several_rounds_before_confirmation and test_dispatch_commit_multi_round_gap_matches_same_round_confirmation (src/scheduler/testsuite/test_scheduler_dispatch_commit.c, task 5.1) prove resInUse survives a 5-round gap with no confirmation event, and match a same-round-confirmation baseline; RED-phase independently verified via a real revert/rebuild/rerun of the fix by both the task 5.1 implementer and its reviewer. Full HA soak (sequential + chaotic, 10 rounds each) confirms no new job loss or divergence, task 5.4.
Verification
Not recorded (migrated from buglog.json).
References
- Related: bug-1611, bug-1613