Patches on contrib projects fail testing with "Drupal installation failed".
In the review log: (during execution of "drush si")
PHP Fatal error: Cannot redeclare system_requirements() (previously declared in /var/lib/drupaltestbot/sites/default/files/checkout/modules/system/system.install:11) in /var/lib/drupaltestbot/sites/default/files/checkout/sites/default/modules/rebuild/tests/fixtures/drupal_sites/dev/modules/system/system.install on line 518
Reported occurrences across various contrib projects so far (up until comment #12 below) :
https://qa.drupal.org/pifr/test/766528
https://qa.drupal.org/pifr/test/230468
https://qa.drupal.org/pifr/test/908703
https://qa.drupal.org/pifr/test/643243
https://qa.drupal.org/pifr/test/908198
https://qa.drupal.org/pifr/test/907473
https://qa.drupal.org/pifr/test/909348
https://qa.drupal.org/pifr/test/909218 on #2463
https://qa.drupal.org/pifr/test/910108 on #2448
All of the above occurred on test clients #2443, #2448 & #2463
Comments
Comment #1
djdevinComment #2
mglamanalso happening here #951004: Allow selecting of multiple media items for a multi value media field in the same dialog.
Comment #3
almaudoh commentedI'm also having this issue at https://qa.drupal.org/pifr/test/230468. At first, I thought it was a random testbot failure, so I made another commit to trigger a test, but still the error persists.
It has stalled tests on smsframework 7.x-1.x branch. Can we fix this?
Comment #4
yched commentedSeen this on scald issue queues too - all patches fail with :
PHP Fatal error: Cannot redeclare system_requirements() (previously declared in /var/lib/drupaltestbot/sites/default/files/checkout/modules/system/system.install:11) in /var/lib/drupaltestbot/sites/default/files/checkout/sites/default/modules/rebuild/tests/fixtures/drupal_sites/dev/modules/system/system.install on line 518
Dunno how many queues that affects, or whether this is only for a couple test clients (all reports so far seem to only involve test clients #2448 & #2463), but this sounds pretty critical :-/
Comment #5
rjacobs commentedI'm seeing this repeatedly in the last couple days as well. Here's one reference: https://qa.drupal.org/pifr/test/908703 (same errors as noted in the comments above). As the error takes place during the installation of the testing instance (before the test-specific tasks are run) it certainly seems like a problem that's independent of the project/patch tested.
I don't mean to be posting another "me too" comment, but I'm hopeful that some patterns from specific examples of failing test results is useful. I can't see from the results which test client was used in my case (maybe I'm missing something obvious within the logs), but there must be some handy info in there?
Comment #6
toddy commentedI can confirm this PHP fatal error during installation of Drupal, at least since yesterday (2014-11-16). The tests ran (after scheduling a retest) on clients #2448 and #2463.
UPDATE: The test also failed on test client #2443.
https://qa.drupal.org/pifr/test/643243
Comment #7
yched commentedCleaned the issue summary a bit and consolidated the reports so far.
Comment #8
toddy commentedAdded test client #2443 to the issue summary.
Comment #9
pieterdcReported https://qa.drupal.org/pifr/test/909348
Comment #10
jonathan1055 commentedAnother failure https://qa.drupal.org/pifr/test/909218 - client #2463
Comment #11
djdevinThere's actually a rebuild module:
https://www.drupal.org/project/rebuild
The latest commit was right before I reported the first failure. Since that module has automated tests, it probably scheduled a new test after the push (Nov 13).
Maybe a test ran for the Rebuild module, and that test client is not cleaning up? That's why only certain test clients are affected?
The module stores a copy of Drupal:
http://cgit.drupalcode.org/rebuild/tree/tests/fixtures/drupal_sites/dev
Comment #12
jonathan1055 commentedThanks djdevin that useful and interesting. How do we go about getting the affected clients cleaned up?
Meanwhile, here is a another just now https://qa.drupal.org/pifr/test/910108 on client #2448
Comment #13
djdevinI can't do anything unfortunately, asked for help in IRC.
Comment #14
isntall commentedI'm going to rebuild the MySQL testbots.
Comment #15
isntall commentedNew bots have been created. Let us know if the issue persists.
Comment #16
rjacobs commentedHummm, it still appears to be an issue. Here's a test result from a re-test I just triggered since comment #15:
https://qa.drupal.org/pifr/test/908703
Looks like that was on client 2508. I can see there there are other recent failures on that client as well (again since comment #15) from other projects. Interestingly though some tests on that same client are working.
My example was a re-test (same test ID). Do I need to try with a brand new patch to test things out properly?
Comment #17
pieterdcI also just triggered a re-test, still fails and also got the same test ID: https://qa.drupal.org/pifr/test/909348
Comment #18
djdevinSame here https://qa.drupal.org/pifr/test/766528
That "Rebuild" module keeps being downloaded. Why? Something seems strange with the dependency build. "Quizz" is not a dependency either. Since "Rebuild" contains a copy of Drupal, maybe the dependency map is attempting to download that. Technically it's a fork of Drupal so it's probably being marked as a valid dependency for a core module. That's what is probably happening in this other case with "Quizz" as the fork contains the same module namespace. Quiz depends on the "filter" module so that's probably pulling "rebuild" from the dependency map, since filter.module exists there.
Somewhat similar to #2378459: Forked module with same .module file breaks testbot which is broken because of a fork.
Comment #19
mpotter commentedAlso seeing it trying to run tests in Features for this issue: https://www.drupal.org/node/2378343
Comment #20
djdevinMakes sense, as features_tests has a hard dependency on "image" - thus bringing in the module that provides it!
Comment #21
brad.bulger commentedSame problem in a new patch test for og_vocab - https://qa.drupal.org/pifr/test/911338
Issue #2379169: empty theme value in og_vocab_element_after_build() causes error
Comment #22
djdevinUpdating title as we know what the issue is now.
Comment #23
isntall commentedI think this issue has been tracked down and resolved.
To recap:
There was the dependency 'rebuild' that satisfied the requirement of getting a copy of core.
There is an array of projects that shouldn't be dependency 'project_dependency_excluded_dependencies' and rebuild was not it.
The rebuild project is now in that array.
Comment #24
rjacobs commentedYep! Two of my previously failing tests have now passed, and the "rebuild" dependency (that was mysteriously added before) is no longer appearing in the related test logs. Looks like the dependency exclusion was the key.
Comment #25
jonathan1055 commentedAll my failing tests have now re-run and passed. Thank you isntall, djdevin and everyone for investigating and fixing.
Comment #26
djdevinThanks, green here too: https://qa.drupal.org/pifr/test/350828
I'd say RTBC
Comment #27
toddy commentedThe tests of my module are passing now, too. Thanks a lot!
https://qa.drupal.org/pifr/test/643243
Comment #28
almaudoh commentedRTBC+1. Re-ran SMS Framework 7.x-1.x-dev branch tests and it passed. Thanks!!
Comment #29
proofoftom commentedIt seems as if all tests are running now as it's been 3 weeks since any reports. Marking issue as fixed. Thanks isntall and everyone else for helping resolve this!