is the CI of mathcomp broken ( failed job)?
that's what I reported here, I don't think Erik has had opportunity to take a look. What has happened is likely that the (Inria) runners that build the Docker containers have failed in some way, maybe @Maxime Dénès could take a look
my error looked more like a nix problem no?
ah sorry, you're right, this looks like a pure Nix problem, specifically with the Cachix caching
There was a report earlier today of a Nix CI problem on MetaCoq, so it looks like this is more general? I didn't look into the failures yet.
looks like we need to upgrade cachix action to v12
https://github.com/coq-community/coq-nix-toolbox/pull/135
not enough :sad:
job failed
Laurent Théry said:
not enough :sad:
job failed
Yes enough, this is an "old job"
Github feels obliged to run both the previous and the new version of the action
Oh and the update is not "automatic", you must run the "updateNixToolBox" then "genNixActions" in the repo
This is both the advantage and drawback of "deterministic" runs
Cyril Cohen said:
Oh and the update is not "automatic", you must run the "updateNixToolBox" then "genNixActions" in the repo
I'm doing it
Not sure what you mean about deterministic, I do rerun failed jobs in my PR on github and it fails again : failed job. How should I trigger it?
https://github.com/math-comp/math-comp/pull/973
ok did a rebase. It seems to go and find cachix :tada:
thanks
Laurent Théry has marked this topic as resolved.
Laurent Théry said:
Not sure what you mean about deterministic, I do rerun failed jobs in my PR on github and it fails again
Well, that sounds pretty deterministic to me :grinning:
@Théo Zimmermann it looks like we have the same cachix issue now in Coq-community (https://github.com/coq-community/reglang/actions/runs/4316263029/jobs/7531802628). What is the recommendation to fix?
Hi @Karl Palmskog I just posted a message with the procedure to follow here
https://coq.zulipchat.com/#narrow/stream/290990-Nix-toolbox-devs-.26-users/topic/Recent.20github.20action.20failures/near/339183907
If you're having trouble please follow-up there.
Jason Gross has marked this topic as unresolved.
I rebased https://github.com/math-comp/math-comp/pull/790 to take advantage of the (hopefully?) fixed CI, but now the CI is not running at all on the PR?
Last updated: Sep 28 2023 at 10:01 UTC