- Issue created by @catch
- First commit to issue fork.
- Merge request !10939Issue #3500489: Performance tests need to run cron β (Closed) created by plopesc
- πͺπΈSpain plopesc Valladolid
Made the suggested change and PerformanceTest is passing again locally.
Since the reported issue is not being reported in the CI jobs, not sure how to test it or whether this one could be merged as it is.
- π¬π§United Kingdom catch
I think it's more important that the tests tests what they're supposed to, than that we test that the tests test what they're supposed to.
We could probably check cron_run_last and make sure it's not 0 - would mean doing an assertion inside ::setUp() but probably we do that elsewhere already.
- π¬π§United Kingdom catch
Or another option here - we could uninstall automated_cron. We don't need it for anything in the tests, we just need it not to run. That definitely doesn't need explicit test coverage then - would need to call ModuleInstaller::uninstall() in ::setUp() where it would otherwise run cron.
- πͺπΈSpain plopesc Valladolid
Uninstalling automated_cron did the trick. It required to update a few cache get counters in a couple of Performance Tests, but I think this is OK now.
It seems that we're doing exactly the opposite of what the IS suggests, though :)
- π¬π§United Kingdom catch
Cache gets are probably the automated_cron configuration.
Updated the issue summary and title. Hoping someone else can RTBC so I can still commit this..
- πΊπΈUnited States nicxvan
Looks good to me, all performance tests have been updated and this should make the queries like up in tests.
- π¨πSwitzerland berdir Switzerland
> Cache gets are probably the automated_cron configuration.
Created π Aggregate cache operations per bin to allow for more specific asserts Active so we have the ability to see this a bit more reliably and can be more specific with assertions when we want to.
- π¬π§United Kingdom catch
A bit concerning that we're getting different results locally vs. gitlab, there's an existing issue about functional js test false negatives on gitlab. However this issue makes sense in itself and is a straight regression from when we removed the cron run in the installer, so let's do it.
Committed/pushed to 11.x, thanks!