- Issue created by @catch
- 🇬🇧United Kingdom catch
Drupal\Tests\taxonomy\Functional\Rest\TermXmlAnonTest 9 passes
Drupal\Tests\taxonomy\Functional\Rest\TermJsonAnonTest 9 passes
Drupal\Tests\taxonomy\Functional\Rest\TermXmlCookieTest 9 passes
Drupal\Tests\taxonomy\Functional\Rest\TermJsonCookieTest 9 passes - Status changed to Needs review
9 months ago 8:17am 4 October 2023 - last update
9 months ago 30,378 pass - @catch opened merge request.
- last update
9 months ago 30,379 pass - last update
9 months ago 30,379 pass - Status changed to RTBC
9 months ago 3:31pm 4 October 2023 - Status changed to Fixed
9 months ago 7:11am 5 October 2023 - 🇬🇧United Kingdom catch
Per #8 it is very tricky to balance. There are broadly two kinds we need to mark slow:
1. The very, very slow tests that absolutely need to start at the beginning of the job, because they are so slow they book-end all the other tests. At the moment the slowest of these are around 6 minutes (down from 13 when we first committed gitlab to core).
2. Moderately slow tests, say 1-2 minutes, that if they happen to start near the end of a test run, will artificially extend the run by those 1-2 minutes, this is 'overhanging' as opposed to book-ending.
However 1 + 2 has to be <= concurrency * parallel jobs, otherwise some of the slowest tests could be pushed out of the first 24 (or whatever concurrency is set at) jobs to run.
Automatically closed - issue fixed for 2 weeks with no activity.