-
Bug
-
Resolution: Unresolved
-
Medium
-
None
-
None
-
None
-
3
-
9223372036854775807
Autotest reports suite FAIL without a failed subtest even when Auster and the suite logs show success. It then starts a subsequent run and terminates it with SIGTERM well before its timeout deadline. The evidence points to Autotest monitoring or result collection; the exact infrastructure defect is not established.
On change 67956 patchset 4, revision 50b8196663bdd3e9aa0e120a628309c8d2d53c04:
- https://testing.whamcloud.com/test_sessions/b384d68f-3ec5-43f6-b96b-ea304d271e8a (el9.8, review-dne-part-6): ost-pools passed every subtest and finished cleanup at 14:40:59. Auster finished with rc 0 at 14:42:46. The execution log says Auster exited unexpectedly at 14:44:05, while still attributing the run to replay-single. It processes the old ost-pools starts around 15:00 and reports Auster Completed with 0 at 15:00:46.
- https://testing.whamcloud.com/test_sessions/cee94f5e-f35e-4b75-acb4-8134f71b69a6 (el9.8, review-dne-part-7): the earlier Auster run finished with rc 0. A subsequent sanity-ec run was terminated with exit 143 at 15:01:26, one second after test_62c started. Its recorded timeout deadline was 16:31:21. Both a synthetic suite FAIL and test_62c TIMEOUT were recorded.
- https://testing.whamcloud.com/test_sessions/ec41ab05-e966-4d76-8eac-966aca50862c (el9.8, review-dne-part-8): Auster finished with rc 0 at 14:26:41, but the execution log says it exited unexpectedly at 14:28:07. The subsequent replay-dual test_12 was terminated with exit 143 at 15:01:31, eleven seconds after the monitor recorded its start and before its 16:31:20 deadline. Both a synthetic suite FAIL and test_12 TIMEOUT were recorded.
- https://testing.whamcloud.com/test_sessions/f3ee396c-93b0-49fb-9396-b0940f153c98 (el10.1, review-dne-part-8): every replay-dual subtest passed and cleanup finished. Auster finished with rc 0 at 14:39:37, yet the monitor reported an unexpected exit. The separate node-reset reboot failure is LU-18696.
The same infrastructure symptom occurs on unrelated change 69392 patchset 2, session https://testing.whamcloud.com/test_sessions/711c6225-a7ef-43c2-a6e8-0812ca737991: Auster completed with rc 0, but its monitor reported Auster exited unexpectedly at 14:31:23. This establishes the defect outside the llog series. Neither change 67956 nor its parent 67532 changes Auster or Autotest. No kernel crash signature or OOM was found in the targeted console and syslog checks.
Rainshadow lists 33 ost-pools suite-level failures in 30 days, including 18 branch failures; that generic category includes other causes and is not a measured rate for this defect. The sanity-ec 62c and replay-dual 12 timeout signatures each occurred once, on this patchset. No local reproduction was attempted because the logs establish premature external termination rather than a test timeout. Incidents INC-2, INC-3 and INC-4 record the infrastructure evidence and are suspected, not active.
JIRA searches for unexpected Auster exits, Autotest monitoring, SIGTERM and incorrect exit status found no matching ticket. LU-19103 concerns an initialization mount error; LU-15152 fixed Auster status propagation from failed suites, whereas these suites passed.
Raised by Patch Watcher run pw-engineer-67956-ps4-26f51b5573ba on https://review.whamcloud.com/67956