Support for Drupal 7 is ending on 5 January 2025—it’s time to migrate to Drupal 10! Learn about the many benefits of Drupal 10 and find migration tools in our resource center.
Problem/Motivation
Random fail in Drupal\page_cache\Tests\PageCacheTest:: testPageCacheAnonymous403404()
fail: [Other] Line 386 of core/modules/page_cache/src/Tests/PageCacheTest.php:
The cache entry expiry time uses the cache_ttl_4xx setting.
See https://www.drupal.org/pift-ci-job/583247
https://www.drupal.org/pift-ci-job/584350
Proposed resolution
Remaining tasks
User interface changes
API changes
Data model changes
Problem/Motivation
Proposed resolution
Remaining tasks
User interface changes
API changes
Data model changes
Comment | File | Size | Author |
---|---|---|---|
#18 | 2847268-18.patch | 984 bytes | tacituseu |
#15 | 2847268-15.patch | 922 bytes | tacituseu |
#4 | 2847268-4.patch | 664 bytes | alexpott |
Comments
Comment #2
alexpottComment #3
alexpottThis could easily be a timing issue...
Comment #4
alexpottLet's add some info to the assertion message to help us.
Comment #5
pwolanin CreditAttribution: pwolanin as a volunteer and at SciShield commentedSporadic fail here: https://www.drupal.org/pift-ci-job/583247
Comment #9
catchCommitted the better assertion message to aid debugging.
Comment #10
tacituseu CreditAttribution: tacituseu commentedhttps://www.drupal.org/pift-ci-job/584350
$difference = $cache_item->expire - (int) $cache_item->created; // (= 3594)
Comment #11
dawehnerWe could increase this offset from 5 to 15 seconds. The actual page cache system is quite unit tested in
\Drupal\page_cache\StackMiddleware\PageCache
so we don't really necessarily loose a lot of test coverage with doing so. Increasing the number to 15 seconds would dramatically decrease the probability for random failures.Comment #12
xjmAdding the HEAD failure to the summary to demonstrate it's not just an issue with a patch.
Comment #13
tacituseu CreditAttribution: tacituseu commentedhttps://www.drupal.org/pift-ci-job/584500
$difference = 3594;
Comment #14
dawehner@tacituseu
I guess 6 seems basically the maximum we get in random test failures, so maybe increasing the different to 10 would already be enough?
Comment #15
tacituseu CreditAttribution: tacituseu commentedCouldn't find more instances.
Comment #16
dawehner10 seconds work for me
Comment #18
tacituseu CreditAttribution: tacituseu commentedPatch against HEAD this time ;)
Comment #20
tacituseu CreditAttribution: tacituseu commentedJust a re-roll so back to RTBC per #16.
BTW in https://www.drupal.org/pift-ci-job/585026 patch failed to apply, yet all the tests ran anyway, seen it a couple of times now.
Also reasoning behind 10 seconds is: 6-7 is extreme case and doesn't occur often, caused by high concurrency on testbot, if it gets any slower would be good to know/trip anyway.
Comment #21
tacituseu CreditAttribution: tacituseu commentedhttps://www.drupal.org/pift-ci-job/584529
The cache entry expiry time uses the cache_ttl_4xx setting. Expire: 1485653810 Created: 1485650215.513
$difference = 3595;
Could someone re-trigger default branch for 8.4.x-dev ? It's stalling issues for more than a day now.
https://www.drupal.org/node/3060/qa
Comment #24
catchCommitted/pushed to 8.4.x and 8.3.x.
Doesn't apply to 8.2.x but since it's test-only, I think we can leave this at 8.3.x.
Comment #25
tacituseu CreditAttribution: tacituseu commented#15 has patch that applies to 8.2.x
Comment #26
xjmComment #28
tacituseu CreditAttribution: tacituseu commentedWell..
https://www.drupal.org/pift-ci-job/600319
$difference = 3590