55 lines
3.4 KiB
Markdown
55 lines
3.4 KiB
Markdown
---
|
|
id: "bug-1940"
|
|
status: "fixed"
|
|
status_updated: "2026-10-10"
|
|
status_note: ""
|
|
component: "apiserver"
|
|
severity: "wrong-result"
|
|
timestamp: "2026-10-10T08:14:46Z"
|
|
last_seen: "2026-10-10T08:14:46Z"
|
|
occurrences: 1
|
|
files: ["src/frontend/apiserver/controllers/jobs.go", "src/frontend/apiserver/backend/clib.go"]
|
|
tags: ["apiserver", "job-control", "signal"]
|
|
related: []
|
|
fix_commits: ["4d2e80c4"]
|
|
gitea_issue: 371
|
|
gitea_synced_status: "fixed"
|
|
gitea_synced_hash: "6b5ca38c13bd"
|
|
---
|
|
|
|
# bug-1940: job control ignores the requested signal and the force-kill flag (apiserver to controller)
|
|
|
|
## 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 control handler passes a JSON event string as content (`controllers/jobs.go:2260-2277`); `signalNumber()` looks it up as a signal name, fails and falls back to TERM/TSTP/CONT; `backend/clib.go:697-705` always sends `sig.force = 0` and `ckpt.kill = 0`, dropping ForceKill and the checkpoint kill/method options (`:746-762`).
|
|
|
|
**Expected:** A kill with a user-chosen signal sends that signal; force kill and checkpoint options reach the controller.
|
|
**Impact:** Wrong result: users cannot send a specific signal or force-kill a job.
|
|
**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. `POST /xce/v1/jobs/<id>/control` with action kill, signal `SIGUSR1`, force_kill true.
|
|
2. The controller receives TERM with force 0.
|
|
- 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
|
|
`ControlJob` (`backend/clib.go`) treated its content (a JSON event) as a signal name, so the lookup always missed and fell back to TERM/TSTP/CONT; it also hard-coded `force = 0`, `ckpt.kill = 0` and never set the checkpoint method. Pre-existing.
|
|
|
|
## Fix
|
|
`backend/convert_pure.go`: new pure `jobCtrlParams()` parses the JSON event or a bare signal name/number, validates it against `JobSignalMapTable` (error instead of fallback), maps force_kill to the signal's force (kill) or the checkpoint's kill flag, and passes the method; `backend/clib.go` fills the C structs from it. Worker-side substitution of the delivered signal was a separate defect, bug-1954.
|
|
|
|
## Verification
|
|
- Test: `TestJobCtrlParams_*` (`backend/jobctrl_params_test.go`): RED `jobctrl_params_test.go:34: kill USR1 force_kill: got signo 15 force 0, want signo 10 force 1` and the checkpoint case, GREEN after (6 tests).
|
|
- 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: `4d2e80c4` (on `scheduler-framework`; worker-branch original in the merge)
|
|
- Review: `.kiro/reviews/project-gap-review-2026-10-10.md`
|