CI on UNIBO premises - #97
Conversation
a4729a3 to
120cace
Compare
c25167d to
df699a7
Compare
|
I could not test the push from GitHub because GitHub actions are currently broken, but the pipeline seems to be now working fine on GitLab side. Needs a bit of cleanup though (squashing a few "update nonfree" commits in particular) |
|
Currently failing only in elab (looks like a legitimate problem on our servers, to be checked) |
df699a7 to
7e1bfaf
Compare
This commit switches magia to use unibo-based infrastructure instead of iis-git.
7e1bfaf to
3cd8c5f
Compare
|
I updated the nonfree commit, fixing the elaboration issue and improving the CI script. |
Unfortunately 6h is a hard cap for GitHub actions, so at the moment there's no real solution for the timeout failing. Re-triggering the action solves the problem since it simply checks the status of the offloaded pipeline for the commit that triggered the CI. |
All GitHub-hosted runners for actions have a maximum timeout limit of 6h (https://docs.github.com/en/actions/reference/limits) |
|
Not much to do about the timeout (unless we change the CI to use Verilator later, which is both faster and potentially more parallel). I'll merge. |
This PR moves CI to local UNIBO premises in place of shared IIS runners.
It also removes regressions running on ri5cy.