It seems that gitlab ci does not start on my PRs to coq/coq or intermediate PR commits.
Also, I'm not able to restart it from the GUI. I see a cancel button, but it fails.
(An example is this pipeline: https://gitlab.com/coq/coq/-/pipelines/224243299 )
-In the short term: could someone start this pipeline and explain to me how I can distinguish between a
real fail and me just loosing in
will-my-pipeline-start-lottery? It's quite frustrating.
-In the long term: Is there a way to improve the experience when opening PRs so that Icontributors do not have to guess if my PR is really broken or CI on gitlab did just not run properly?
Look at the red flag. It states "pipeline job activity limit exceeded". There is nothing you can do right now. No point in restarting it, because the limit will still be exceeded.
Do I need to intervene at some point or is it scheduled to start once space is free?
or is a quota reached and there is no money left?
I never had to restart to it, so I tend to think it is the latter, i.e., auto start.
There is some silly bug, and if you cancel + restart the pipeline is actually run.
No point in restarting it, because the limit will still be exceeded.
That's actually not true, probably due to a GitLab bug.
As Enrico said, you can do the cancel + restart danse on GitLab (but for this you need to have the access right in the GitLab org).
sudo gitlab run pipeline
Or you can click on "Re-run" in the GitHub Checks tab (for this you only need to have write access in the GitHub repo).
@Fabian Kunze I have just invited you to join our
@coq/contributors team, which means that as soon as you accept the invitation, you will have write-access in the GitHub repo.
The usual warnings: while our main branches are protected, there is nothing preventing you to push a branch to the repo now. Please don't do it. Be careful to always push branches to your fork.
thanks, I called the repo "upstream" to avoid those things :)
@Théo Zimmermann does osx CI save the test suite logs?
as we do on unix?
Not as artifacts, no. But it prints them out in the log IIRC.
I have a silly failure,
grep related, here: https://github.com/coq/coq/pull/13523/files
Fabian Kunze said:
thanks, I called the repo "upstream" to avoid those things :smile:
Good strategy (this is mine too)!
@Fabian Kunze what was the drill about grep then?
it was not grep, but it was that i depended on make-internals regarding the order of recipes. (and on mac, they where resolved in another order)
i changed that
So, which is it? Does the pipeline start automatically? Or is there a good samaritan who always restarts mine?
ok, good for you. here I'm still lost
I often do restart pipelines from other people.
And I'm not the only one.
I do restart all the ones I see stuck
And pipelines do get started automatically when activity levels are not so high of course.
bein doing that for the last 2 weeks (release frenziness)
I'll try to find some time to have coqbot automatically cancel outdated pipelines and restart the ones that get stuck.
I'm reluctant to do the latter without the former, because if we go well over the limit, it's likely to become suspicious to GitLab.
Last updated: Jun 11 2023 at 00:30 UTC