-
Notifications
You must be signed in to change notification settings - Fork 157
Description
While executing a test, the test crashed the machine.
tmt detected test timeout:
23:51:11 out: stress-ng: info: [237771] dispatching hogs: 12 schedpolicy
00:44:37 Command returned '124' (timeout).
00:44:37 Test duration '1h' exceeded.
The the results in the end was still pending.
- name: /kernel-tests/stress/stress-ng
result: pending
original-result: pending
<testcase name="/kernel-tests/stress/stress-ng" result="pending">
https://artifacts.osci.redhat.com/testing-farm/cfda10c9-0d19-428d-8c6a-5c1ecfe222dc/results.xml
Edit by @happz: the important thing here is the test killed the guest. Without this, the timeout would be reflected, but because the guest was dead (watchdog was not enabled), the post-test attempt to pull workdir from the guest crashed, which led to exceptions. The timeout was, therefore, not reflected in the results. Should the guest remain available, timouet would eventually replace the pending
outcome.
I believe this error state should be recorded in results first before moving on to fetching the workdir. tmt would still crash because the guest remains dead, but the test result would be correct.
Metadata
Metadata
Assignees
Labels
Type
Projects
Status