-
Notifications
You must be signed in to change notification settings - Fork 17
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
CI Fixes #432
CI Fixes #432
Conversation
@@ -179,7 +179,7 @@ jobs: | |||
regressions: | |||
needs: process-results | |||
# Always check for regressions, as this can be skipped even if an indirect dependency fails (like a test run) | |||
if: always() | |||
if: always() && github.event_name != 'pull_request' |
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.
I think that for the time being we should disable these checks on PRs. There is just too much variability to trust single measurements.
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.
Other option would be to push the number of standard deviations to 4 (currently at 3) so we will only see reports for crazy spikes.
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.
We already see huge spikes (well beyond any reasonable number of standard deviations). They aren't especially common, but if we run 100 tests and there is a one percent chance of a spike, that means a red CI for PRs most of the time.
I still don't think we have a good understanding of why the spikes happen, but I don't think checking for regressions on PRs is reliable enough due to them.
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.
Thanks @ian-r-rose! It looks like reducing the number pytest-xdist
workers fixes things. Mind pushing an empty commit after the current CI run is done, just to confirm things are consistently green?
Attempt to fix #404