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>
4 KiB
| id | status | status_updated | status_note | component | severity | timestamp | last_seen | occurrences | files | tags | related | legacy | ||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| bug-1780 | fixed | 2026-09-24 | apiserver | unknown | 2026-09-07T07:55:00Z | 2026-09-19T09:00:00Z | 1 |
|
|
true |
bug-1780: User-reported (xcecli): 'xcecli jobs submit -n 2 --env /run/current-system/sw/bin/sleep 100'…
Description
User-reported (xcecli): 'xcecli jobs submit -n 2 --env /run/current-system/sw/bin/sleep 100' submits successfully but the job immediately fails with Status . xcecd.log shows: '[WARN][controller_connector.c:822] No execuser provided in startjob request, defaulting to nobody', then 'Starting worker as user nobody to process command...', then 'Worker 301734 exited with code 126'.
Environment
Not recorded (migrated from buglog.json).
Reproduction
Not recorded (migrated from buglog.json).
Root cause
modelJobToSubmit() (src/frontend/apiserver/backend/convert.go, converts a Go models.JobInfo into the C job_startinfo_t sent to the controller) populated si.cmdline/launch_type/job_pre_exec/job_post_exec but never set si.execuser, even though job_startinfo_t (src/include/xce/types.h) has an execuser field and models.JobInfo already has an ExecUser field (used only on the read/display path, in convert_pure.go's rawJobToModel). CreateJob() (controllers/jobs.go) also never populated JobInfo.ExecUser when building the job to submit -- there is no --exec-user flag on xcecli either, so the field was simply always empty at submission. The controller (src/controller/job.c) forwards job.submit->start_info verbatim to the worker without ever touching execuser. xcecd's controller_connector.c (lines 817-822) defaults an empty/NULL execuser to the literal Linux user 'nobody', then (lines 1037-1040) copies that into job_ctx->limits->user_name, which xcecd_dbus_manager.c uses as systemd-run's User= for the job's transient unit. 'nobody' has no home directory/login shell and typically cannot execute or access the real submitting user's binaries/cwd, so the transient unit fails with 'Permission denied' -- surfacing to xcecd_worker_utils.c as exit_code=126, which xcecli then reports as job Status . Every job submitted via xcecli/apiserver was silently running as 'nobody' regardless of who submitted it.
Fix
FIXED (2026-09-07), TDD RED->GREEN. Added resolveExecUser(execUser, submitUser string) string to convert_pure.go -- a pure-Go helper (no CGO types) that returns the caller-specified exec user if present, else defaults to the submitting user; this mirrors the existing pure/CGO split in this file (rawJobToModel etc. already keep pure field-mapping logic separate from the CGO struct-building in convert.go specifically so it's testable without CGO -- and because a _test.go file can never itself import "C" (golang/go#13938), this split is the only way to unit-test any part of modelJobToSubmit's logic). modelJobToSubmit() now sets si.execuser = C.CString(resolveExecUser(job.ExecUser, job.SubmitUser)); freeJobSubmit() now also frees si.execuser (previously would have leaked once set). New tests in convert_test.go: TestResolveExecUser_DefaultsToSubmitUserWhenEmpty, TestResolveExecUser_PrefersExplicitExecUser, TestResolveExecUser_EmptyWhenBothEmpty -- pass under CGO_ENABLED=0 go test ./backend/... (RED confirmed by the function not existing before this change). Verified with a full /xce-build apiserver asan --incremental: exit 0, all frontend Go suites pass including apiserver/backend (7.595s, real CGO build), fresh xce_apiserver/xcecli binaries produced. Live re-run of the exact reported repro (xcecli jobs submit ... /run/current-system/sw/bin/sleep 100) not yet performed in this session -- recommend a live smoke test to confirm the job now runs as the submitting user and reaches DONE/RUN instead of EXIT before closing.
Verification
Not recorded (migrated from buglog.json).
References
Not recorded (migrated from buglog.json).