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>
3.2 KiB
| id | status | status_updated | status_note | component | severity | timestamp | last_seen | occurrences | files | tags | related | fix_commits | legacy | ||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| bug-1519 | fixed | 2026-09-24 | b6a2f5ec (bug-1756) |
worker | unknown | 2026-08-11T12:49:17.815Z | 2026-09-19T09:00:00Z | 2 |
|
|
|
true |
bug-1519: ==ERROR: AddressSanitizer: Joining already joined thread, aborting
Description
==ERROR: AddressSanitizer: Joining already joined thread, aborting.
Environment
Not recorded (migrated from buglog.json).
Reproduction
Not recorded (migrated from buglog.json).
Root cause
UNRESOLVED / not root-caused. xcecd's root check (main.c:288 geteuid()!=0) was bypassed for sandbox testing via unshare -r (user-namespace fake root, euid=0 but not real root). Startup then immediately hit an ASAN double-pthread_join abort during/after the HA-cluster-config init log line (main.c:435), before any further logging. Never reproduced under real root -- unknown whether this is a genuine double-join bug in xcecd's daemonize/thread-spawn path (e.g. duplicate cleanup on an early-init error branch) or an artifact of unshare's user-namespace semantics (e.g. a syscall xcecd's threading depends on behaving differently under CLONE_NEWUSER, such as thread creation ordering or signal delivery). Blocked further component-failover-adoption Task 3.3 Item-3 E2E validation from getting a live worker registered, which in turn meant no real job ever produced a scheduler decision (picker/allocator never had a matching candidate to dispatch to or explicitly PEND against), so the scheduler's decision-publish/rejoin code path (task 3.2) could not be exercised via a fully live job in this sandbox.
Fix
Not fixed -- out of scope for Task 3.3 (which validates scheduler HA reconnect, not worker daemon startup). Worked around by validating scheduler join/failover/reconnect behavior via direct socket-state inspection (ss/proc fd) instead of a full live job pipeline; the decision-publish retry+rejoin logic itself already has white-box unit coverage from task 3.2 (test_scheduler_decision_publish_retry.c) and task 3.3 item 2 (test_scheduler_backpressure.c). If a future session needs a live worker in this sandbox, retry under real root (not unshare -r) first to isolate whether this is userns-specific. UPDATE (2026-08-20, build-time-versioning task 3.3 review): independently reproduced under REAL sudo root (not unshare -r) -- same signature, same timing (fires during later thread-join startup work, strictly AFTER the HA-cluster-config INFO log line, both with and without --peers). This refines/answers the prior 'never reproduced under real root' open question: it is NOT a userns/unshare-specific artifact -- the abort reproduces identically under genuine root. Still not root-caused/fixed; still does not block this session's version-display changes (main.c's new --version/log-line code runs entirely before the point where this abort fires).
Verification
Not recorded (migrated from buglog.json).
References
Not recorded (migrated from buglog.json).