In Drupal\system\Tests\Update\UpdatePathRC1TestBaseFilledTest::testUpdatedSite running updates is not being executed. If we run the updates on the current head without any changes everything should be fine, but unfortunatelly it is not and the test for checking that the entity schema is up to date fails.
This one is not causing any data loss but something looks wrong in the head, so I am prioritizing it as critical until it is clear what is wrong. I guess something got in after RC1 was out without checking for entity schema updates.
| Comment | File | Size | Author |
|---|---|---|---|
| #17 | entity-schema_updates_rc1-2607364-17.patch | 858.54 KB | plach |
| #16 | entity-schema_updates_rc1-2607364-15.patch | 865.41 KB | plach |
Comments
Comment #2
hchonovDefining as blocker, because of this an rc target triage issue #1810394: Site configuration with domain based language negotiation results in redirecting authenticated users to a different domain when accessing a content entity route for translation language different from the interface language is failing.
Comment #4
hchonovThe patch from #2602996: Upgrade path for #2597628 solves the issue here.
Comment #5
alexpottOne thing that is really interesting is that
Drupal\system\Tests\Update\UpdatePathRC1TestBaseFilledTestis failing butDrupal\system\Tests\Update\UpdatePathTestBaseFilledTestis not.Comment #6
catchIs it possible that this is because the RC1 update is actually beta12 -> RC1 -> RC2 - which would be exactly the head2head scenario that #2602996: Upgrade path for #2597628 fixes.
#2600086: Create a script to generate a filled database dump and use it to replace the pseudo-RC1 one we currently have would be useful.
Comment #7
plachProbably. Let's see :)
Comment #8
plachBetter patch
Comment #10
catchIn that case I think we should promote to #2600086: Create a script to generate a filled database dump and use it to replace the pseudo-RC1 one we currently have to critical - since our upgrade tests don't test the actually-supported upgrade path.
And the fix for pre-RC1 databases can stay in head2head.
Comment #11
effulgentsia commentedPromoting #2600086: Create a script to generate a filled database dump and use it to replace the pseudo-RC1 one we currently have to Critical might be overkill for this issue, but a good idea for its own reasons.
If we want to resolve this issue prior to that one, can we simply fix
drupal-8-rc1.filled.standard.php.gzby taking what's there now, running #8 or similar on it, and re-exporting that?Comment #12
catchLet's retitle for the actual bug.
I'm neutral on #11, it means we could run the test without failures, but it makes that database even more of a monster.
Comment #13
plachI think we should either go with #11 or close this as a duplicate of #2600086: Create a script to generate a filled database dump and use it to replace the pseudo-RC1 one we currently have and make it critical.
Comment #14
plachPersonally I'd be fine with an incremental fix.
It's just matter of "dropping" a couple of inserts, I think, no need to run #8 actually.
Comment #15
plachTrying #14.
Comment #16
plachHere it is
Comment #17
plachI forgot to rebase before rolling the full patch, the review patch in #16 is correct.
Comment #19
catchOK, let's get this in since it's an improvement on the status quo.
Comment #20
alexpottCommitted c4149af and pushed to 8.0.x. Thanks!