Fix ci #80
Fix ci #80
Conversation
jacomago
force-pushed
the
fix-ci
branch
5 times, most recently
from
7eeb4a8 to
90ce57e
Compare
|
@jacomago could you describe your merge request in more specific way in your comment? This is not the personal repository, and not the ESS specific repository. All other facilities should know what you are trying to do through this merge. We don't have the template, but you can find another pull request as your example which Michael did in this repository. |
|
@mdavidsaver Do you know why the macos12 ones are flaky? They seem to pass on reruns... |
I have probably encoded some Linux specific assumption into that test. I don't immediately see what though. 100 second really should be long enough for even the slowest GHA runner. It might be helpful for someone with a mac to try running this test locally. |
|
@mdavidsaver Locally I'm getting a failure on 100 seconds but not on 110, and flaky inbetween. Shall I just bump the number? I don't know if this is to do with cpu seconds vs real time seconds... |
|
|
Regarding the mac os test flakiness, this seems to have been a long-standing issue. Even going back to 1.0 we see that the socket timeout test for mac os just... fails some of the time. It might be better to make that test be skipped on mac os... |
|
|
Out of curiosity, why split up the linux/mac/windows tests? Those are doing the exact same things each, aren't they? |
|
Mac: different compiler, no static builds, no cross builds. On GitHub Actions, availability of runners may differ a lot by OS. You get a better overall timing by parallelizing the jobs that usually wait longer. ("Start waiting now.") |
|
My question was more about the matrix used for each case; the steps are the same, but they simply use a different matrix. However, looking a bit more carefully it seems that it isn't an easy fix; it's sort of like several "block diagonal" matrices (to abuse mathematical terms) and it isn't really easy to split that up without some possible yaml anchoring or something similar. Maybe this is best and cleanest. |
| - name: Linux centos 7 | ||
| image: centos:7 | ||
| cmp: gcc | ||
| configuration: default | ||
| base: "7.0" |
There was a problem hiding this comment.
I see that we are removing centos7 support from CI?
There was a problem hiding this comment.
They have.
The only way to run CentOS7 is in self-maintained containers.
There was a problem hiding this comment.
I thought that might be the case, but it was not mentioned in the pull request.
|
@mdavidsaver or @ralphlange review? |
|
Looks good to me. (Note that - in other repos - I started adding CentOS-8 and Rocky-9 docker builds instead of the CentOS-7 that won't work on GHA anymore. No hurry - just saying.) |
Update the client ci scripts to align with the recsync client submodule ci.
Make sure all the builds pass