Conversation
Since HPE rdma tests are quickest (only 15-20 min) if they fail, then fail the entire workflow run, all jobs to save some free tier runners
mhae
left a comment
mhae
left a comment
There was a problem hiding this comment.
Actually, I remembered that this probably not going to work.
See https://github.com/orgs/community/discussions/26311
I tried implementing the cancel early on when I did the runners and didn't get it to work.
we cancelling the jobs like this today, see https://github.com/spdk/spdk-ci/blob/main/.github/scripts/parse_gerrit_webhook.sh |
| with: | ||
| path: ./output | ||
| name: hpe-job-nvmf-rdma | ||
| - name: Cancel all jobs enitre workflow run if this job failed |
There was a problem hiding this comment.
Shouldn't the same be added to common tests workflow? In this one common tests part failed after a minute or so, but HPE runner was still busy anyway.
There was a problem hiding this comment.
It rarely happens. Since how runner only executes a single test. It was true only when how was offline for a bit…
There was a problem hiding this comment.
Rarely, but still - why keep it busy in such a case? Honestly it was the first failed build I clicked on, so maybe it happens not as rarely :)
3 participants
Since HPE rdma tests are quickest (only 15-20 min) if they fail, then fail the entire workflow run, all jobs
to save some free tier runners