[00:49:13] <MirahezeRelay>	 <posix_memalign> Is there a procedure of checks before updating extensions on prod? https://issue-tracker.miraheze.org/T12837 is still reproducible, so it just needs an update to have https://gerrit.wikimedia.org/r/c/mediawiki/extensions/TimedMediaHandler/+/1200001 applied.
[00:58:34] <MirahezeRelay>	 <abaddriverlol> I think you have to upgrade vendor to fix this since the extension is not in <https://github.com/miraheze/mediawiki-repos/blob/main/composer.local.json>
[00:59:32] <MirahezeRelay>	 <abaddriverlol> I'm not sure rn whether there's release branches for vendor, let me check
[01:02:05] <MirahezeRelay>	 <abaddriverlol> Actually it looks like the dependency is actually in extensions/TimedMediaHandler/vendor for some reason so upgrading just the extension should be fine
[01:02:36] <MirahezeRelay>	 <abaddriverlol> Regarding checks, I usually test the upgrade on beta before deploying to prod, just to make sure nothing explodes
[01:08:35] <MirahezeRelay>	 <posix_memalign> Great. Thanks! Looks like Beta already has the patch applied thanks to being on a newer branch.
[01:11:23] <MirahezeRelay>	 <abaddriverlol> [1/2] The dependency isn't up to date yet on beta, at least on 1.44
[01:11:23] <MirahezeRelay>	 <abaddriverlol> [2/2] https://cdn.discordapp.com/attachments/1006789349498699827/1464789716393263246/image.png?ex=6976bf3a&is=69756dba&hm=b77114300fa85178e29e299edaaa42f86b9d4aecbc8edd3ddae7e07465174470&
[01:11:58] <MirahezeRelay>	 <abaddriverlol> Not sure whether there's another 1.44 wiki on beta, otherwise you can use <https://morerandomstuff.mirabeta.org/wiki/Special:Version> to test
[01:14:57] <MirahezeRelay>	 <abaddriverlol> [1/5] ```sh
[01:14:58] <MirahezeRelay>	 <abaddriverlol> [2/5] [somerandomdeveloper@test151:~]$ cat /srv/mediawiki/1.44/extensions/TimedMediaHandler/composer.json | grep getid3
[01:14:58] <MirahezeRelay>	 <abaddriverlol> [3/5]                 "james-heinrich/getid3": "~1.9.24"
[01:14:58] <MirahezeRelay>	 <abaddriverlol> [4/5] ```
[01:14:59] <MirahezeRelay>	 <abaddriverlol> [5/5] I think what happened is that I deployed a security upgrade with `--world`, which pulled the TMH repo but didn't run composer. So upgrading the extension should in theory update the dependency this time
[01:17:09] <MirahezeRelay>	 <abaddriverlol> Or that is my assumption at least, I'm not sure whether mwdeploy actually automatically runs composer if `composer` is set to `true` in the mediawiki-repos entry
[01:17:15] <MirahezeRelay>	 <posix_memalign> Oh that's good to know, though I'm dealing with another issue right now. I'm seeing `You can only upload PCM (Pulse Code Modulation) WAV.` on exttest which makes no sense because running getid3 locally tells me that the codec is `Pulse Code Modulation (PCM)`.
[01:17:47] <MirahezeRelay>	 <abaddriverlol> Oh
[01:28:42] <MirahezeRelay>	 <abaddriverlol> I think if upgrading didn't work (which should probably be fixed in mwdeploy), you could run `sudo -u www-data composer install --no-dev` in the extension staging folder manually
[01:30:11] <MirahezeRelay>	 <abaddriverlol> Puppet is supposed to run that in theory <https://github.com/miraheze/puppet/blob/5dd162c9926d3111a0494cf55df09b2b96bec8f1/modules/mediawiki/manifests/extensionsetup.pp#L80>
[01:36:55] <MirahezeRelay>	 <cosmicalpha> You can delete the extension folder in staging and run puppet then deploy with mwdeploy.
[01:38:04] <MirahezeRelay>	 <abaddriverlol> Yep
[01:38:44] <MirahezeRelay>	 <abaddriverlol> Maybe we should document the ways to upgrade various dependencies (e.g. vendor or ones like this) at some point
[13:49:49] <MirahezeRelay>	 <thewwrnerdguy> https://kraftwerk.miraheze.org/ what's going on here
[13:54:08] <MirahezeRelay>	 <blankeclair> i suspect slot_roles and content_models go kaput
[13:54:21] <MirahezeRelay>	 <blankeclair> i'd fix, but laptop dead + i still haven't bothered to reenable shell
[17:58:59] <MirahezeRelay>	 <posix_memalign, replying to thewwrnerdguy> [1/10] Excerpt (basically what Claire said):
[17:58:59] <MirahezeRelay>	 <posix_memalign, replying to thewwrnerdguy> [2/10] ```
[17:58:59] <MirahezeRelay>	 <posix_memalign, replying to thewwrnerdguy> [3/10] Main slot of revision not found in database. See T212428.
[17:59:00] <MirahezeRelay>	 <posix_memalign, replying to thewwrnerdguy> [4/10] from /srv/mediawiki/1.44/includes/Revision/RevisionStore.php(1535)
[17:59:00] <MirahezeRelay>	 <posix_memalign, replying to thewwrnerdguy> [5/10] #0 /srv/mediawiki/1.44/includes/Revision/RevisionStore.php(2101): MediaWiki\Revision\RevisionStore->constructSlotRecords(string, array, int, MediaWiki\Title\Title)
[17:59:00] <MirahezeRelay>	 <posix_memalign, replying to thewwrnerdguy> [6/10] #1 [internal function]: MediaWiki\Revision\RevisionStore->MediaWiki\Revision\{closure}(stdClass)
[17:59:01] <MirahezeRelay>	 <posix_memalign, replying to thewwrnerdguy> [7/10] #2 /srv/mediawiki/1.44/includes/Revision/RevisionStore.php(2077): array_map(Closure, array)
[17:59:01] <MirahezeRelay>	 <posix_memalign, replying to thewwrnerdguy> [8/10] #3 /srv/mediawiki/1.44/includes/language/MessageCache.php(589): MediaWiki\Revision\RevisionStore->newRevisionsFromBatch(array, array)
[17:59:01] <MirahezeRelay>	 <posix_memalign, replying to thewwrnerdguy> [9/10] #4 /srv/mediawiki/1.44/includes/language/MessageCache.php(438): MessageCache->loadFromDB(string, null)
[17:59:02] <MirahezeRelay>	 <posix_memalign, replying to thewwrnerdguy> [10/10] ```
[18:01:58] <MirahezeRelay>	 <zppix, replying to posix_memalign> Why can’t all error messages be this helpful
[18:17:50] <MirahezeRelay>	 <posix_memalign, replying to posix_memalign> Should be fixed now with Tech:Fixing_slot_roles_and_content_models
[18:32:26] <MirahezeRelay>	 <posix_memalign, replying to zppix> Including a task number is usually very helpful
[18:37:37] <MirahezeRelay>	 <rhinosf1, replying to zppix> Would make our lives too easy
[21:01:08] <MirahezeRelay>	 <paladox, replying to posix_memalign> problem is why does this happen
[21:01:13] <MirahezeRelay>	 <paladox> shouldn't be happening
[21:01:47] <MirahezeRelay>	 <paladox> it's been happening for years for us but it indicates a problem i guess and the root cause should be fixed
[21:31:25] <MirahezeRelay>	 <pskyechology, replying to paladox> wonder if it could be for similar reasons why creating wikis fails when mwtask181 disk is full
[22:58:35] <MirahezeRelay>	 <notaracham> Oh, that reminds me, we had a batch of weird ones the other day on the 23rd w/ both failed rights assignment and in one case a wiki showing as deleted. 🤔