Open chrismeyersfsu opened 3 months ago
So… Re-runs are made to always fail so that they don't hide flakiness. There's a || exit 1
in the command. That is intentional.
The only problematic bit I noticed are tracebacks coming from commands_post
.
Any other action item I'm missing?
I think I'm wrong about coverage failure being the reason.
I was trying to point out that all tests pass (first screenshot)
Then the re-run seems to run ALL the tests. I would expect it to only re-run the tests that failed above (of which there are none) (second screenshot)
There is also an array index error but I think that may be related to coverage.
No, the index error is related to the last verbose tox call w/o coverage collection. I only tested the first one because there were no failures in the skeleton. So reruns weren't triggered. I fixed that already.
The reruns are generic and trigger on any errors in previous GHA steps.
The pytest rerun is invoked with --lf
to pick up the failed tests but since there's none, all are executed again. I think you're right that coverage likely causes the failure of that first testing invocation.
https://github.com/ansible/awx-plugins/actions/runs/10597353901/job/29367631346?pr=17
FAIL Required test coverage of 100% not reached. Total coverage: 40.78%
It looks like the re-run failed tests with higher verbosity triggers re-running ALL tests when the coverage check fails. In the case above all the tests passed both runs.