[Ops] More resilient test failure collection#175423
[Ops] More resilient test failure collection#175423delanni wants to merge 2 commits intoelastic:mainfrom
Conversation
|
/ci |
💛 Build succeeded, but was flaky
Failed CI StepsMetrics [docs]
To update your PR or re-run it, just comment with: |
|
Pinging @elastic/kibana-operations (Team:Operations) |
jbudz
left a comment
There was a problem hiding this comment.
I'm not opposed to trying it, but are we sure the junit reports are created? It seems we're terminating early due to a few missing awaits causing unhandled promises at
|
You're probably right, the reports are not created at that state. My attempt with this PR was to still add the error annotation, and upload any other artifacts that might have been created during the test runs, but now I see that the tagging logic needs the JUnit/Html reports to produce the annotation from. I'll not merge this, I don't want to mess with the error codes at this point, not to prevent any retries or similar. I think that error was coming from a leakage that I tried to fix here: #175430 |
Summary
See: https://buildkite.com/elastic/kibana-on-merge/builds/40574#018d3ab8-240d-44b2-975a-9cbfa09d3177
If I interpret the results right, some FTR runs fail on the JUnit result collection step, thus not allowing any further steps to run from the post-command actions, this prevents the annotation and screenshot uploading from happening.
This PR adds a more resilient execution of post-command actions, collecting the resulting exit code from several steps before exiting.