- Issue created by @alexpott
- π¬π§United Kingdom alexpott πͺπΊπ
This is related to π package_manager kernel tests are slow Active - think we could do both though.
- π¬π§United Kingdom alexpott πͺπΊπ
This is needs to be looked at again now that π package_manager kernel tests are slow Active has landed.
- π¬π§United Kingdom catch
Rebased and added a couple of build test changes based on https://git.drupalcode.org/project/drupal/-/jobs/3638126
- π¬π§United Kingdom catch
https://git.drupalcode.org/project/drupal/-/jobs/3638126 was about 3.5 minutes.
Added an extra job and more concurrency for build tests because package_manager has more than doubled the linear test time for build tests.
https://git.drupalcode.org/project/drupal/-/jobs/3639334
https://git.drupalcode.org/project/drupal/-/jobs/3639335After that both jobs finish in under 2m20s. With a single job with the CPU request and concurrency double, it took an extra minute for the one test, see https://git.drupalcode.org/project/drupal/-/jobs/3639208
This might be because there's more scope for the two jobs to claim additional unused CPUs? Not sure.
Either way I think this is reviewable now.
- π¬π§United Kingdom catch
OK back down to one job with a higher CPU and 5 concurrency, first run took a long time, second run was 2m30s so that is pretty good. Stopping here. https://git.drupalcode.org/project/drupal/-/jobs/3645012
- πΊπΈUnited States smustgrave
Could the issue summary be updated some? Title and summary seemed to be about marking tests #slow but seems to be doing more now.
- πΊπΈUnited States smustgrave
From looking at https://git.drupalcode.org/project/drupal/-/jobs/3649404 and searching for Drupal\Tests\package_manager\Kernel\ComposerPluginsValidator looks like last one went down to 130seconds which definitely seems like an improvements.
Going to go on a limb and mark it.
- π¬π§United Kingdom catch
Yes it's better.
ComposerPatchesValidator
is still 160s on that run, but this is less than various other tests. Just kicked off a run and none of the package_manager tests were the slowest on that run. -
larowlan β
committed 252c4b8d on 11.x
Issue #3490512 by catch, alexpott, smustgrave: Improve run times of...
-
larowlan β
committed 252c4b8d on 11.x
- π¦πΊAustralia larowlan π¦πΊπ.au GMT+10
Committed to 11.x - thanks folks