-
Notifications
You must be signed in to change notification settings - Fork 94
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Nextest: reactivate completeness #2373
Conversation
They take a lot of memory for now, and make the CI fail from time to time even though the code works. Reviewers from now should run the completeness tests locally. It is a temporary fix to move on.
Limit the number of threads. Can be useful if they take a lot of memory
37c0ff9
to
7eef83e
Compare
7eef83e
to
3217f1d
Compare
|
||
[[profile.ci.overrides]] | ||
filter = "test(completeness)" |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Nextest docs have a section on heavy tests: https://nexte.st/docs/configuration/threads-required/
|
||
[[profile.ci.overrides]] | ||
filter = "test(completeness)" |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Nextest docs have a section on heavy tests: https://nexte.st/docs/configuration/threads-required/
[[profile.ci.overrides]] | ||
filter = "test(completeness)" | ||
threads-required = 2 |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
My prediction would be that it's more optimal to give all the heavy tests all the threads (or at least half), because they're quite parallelisable anyway. But we can do it in the future when we hit the ceiling next time.
In 2372, we deactivate the tests checking the completeness. The reason was that the CI was failing from time to time with a SIGKILL (see GH actions history of 2372 and 2368 - might be others). The suspected reason was that too much memory or too much CPU was used, and the pod/container on the GH actions runner was killed. It would make sense because the completeness tests require to commit, evaluate and interpolate many polynomials.
This PR reactivate the completeness tests, by setting a particular nextest configuration only for them.
It seems that it does increase the CI runtime from 20m to 23m, unfortunately. It would make sense as we request to run on less threads. I think we should try to change some Nextest configuration parameters to see what works better.