mirror of https://github.com/pulumi/pulumi.git
33a5cde5d4
Acceptance tests on Windows sometimes fail for mysterious reasons. It's unclear to me whether this is because the tests sometimes genuinely take too many resources, and break the windows host, or whether it's instability of the windows CI runners in general (it could also be both). This is very annoying during merges, because for any failed test we need to re-do the merge, which means re-running all tests, which is both expensive and slow. Instead, run the acceptance tests on windows during PRs always, but don't run them during merge. I think this is okay to do because if there is a mismerge, I don't expect exclusively windows tests to fail, but I expect that to also affect Linux and MacOS runners. It seems very unlikely for a Windows specific problem to pop up only when multiple PRs are merged together. It is also convenient to run the windows tests during PRs in the first place, because those are tests developers usually don't run on their own machine, so it's not super unlikely they fail. We used to not run these tests on PRs because it's kind of expensive to run them, but we have more headroom now, so running them during PRs should no longer be a big issue. |
||
|---|---|---|
| .. | ||
| ISSUE_TEMPLATE | ||
| scripts | ||
| workflows | ||
| CODEOWNERS | ||