[08:34:54] <wikibugs>	 (03PS1) 10Kosta Harlan: phpbench: Use git-checkout instead of git-switch [integration/quibble] - 10https://gerrit.wikimedia.org/r/774379 (https://phabricator.wikimedia.org/T291549)
[08:46:24] <wikibugs>	 (03CR) 10Hashar: [C: 03+2] phpbench: Use git-checkout instead of git-switch [integration/quibble] - 10https://gerrit.wikimedia.org/r/774379 (https://phabricator.wikimedia.org/T291549) (owner: 10Kosta Harlan)
[08:48:53] <wikibugs>	 (03CR) 10Kosta Harlan: "recheck" [integration/quibble] - 10https://gerrit.wikimedia.org/r/758783 (https://phabricator.wikimedia.org/T235449) (owner: 10Kosta Harlan)
[08:49:29] <kostajh>	 hashar: do you want to put https://gerrit.wikimedia.org/r/c/integration/quibble/+/763333 in the next release or just make a release to fix the phpbench issue only?
[08:49:45] <hashar>	 might as well pill up patches 
[08:49:50] <hashar>	 oh
[08:50:14] <hashar>	 the Spi I had a quick look at it but it generates way too much log
[08:50:18] <hashar>	 that would need a bit of tuning
[08:50:24] <kostajh>	 ah, ok
[08:50:31] <hashar>	 I am going to do the git switch fix
[08:50:41] <hashar>	 but surely we can enhance those logs
[08:50:51] <hashar>	 notably if we had timestamp and the reqId logged that would be great
[08:51:23] <kostajh>	 I think there is a performance boost to using MWLoggerDefaultSpi as well, at least that is what I noticed when switching to it locally
[08:53:04] <wikibugs>	 (03PS5) 10Kosta Harlan: Logging: Use MWLoggerDefaultSpi [integration/quibble] - 10https://gerrit.wikimedia.org/r/763333 (https://phabricator.wikimedia.org/T225730)
[08:53:33] <kostajh>	 timestamp is logged 
[08:53:39] <kostajh>	 as is request ID
[08:54:08] <kostajh>	 I'm not sure it generates more logs than the default... can you point me to an example?
[08:55:24] <hashar>	 irrc we had log from both $wgDebugLog being set and from the Spi config
[08:57:29] <hashar>	 what I am wondering is
[08:58:09] <hashar>	 whether we could move most of those settings to mediawiki/core includes/DevelopmentSettings.php
[08:58:32] <hashar>	 maybe with a conditional based on `CI` environment variable being set
[09:00:28] <hashar>	 	mw-debug-cli.log.gz	Mar 1, 2022, 12:43:17 PM	8.93 MB	 view
[09:00:28] <hashar>	 	mw-www-cli.log	Mar 1, 2022, 12:34:47 PM	553.69 KB	 view
[09:00:28] <hashar>	 	mw-www-debug.log	Mar 1, 2022, 12:37:56 PM	41.46 MB	 view
[09:01:07] <hashar>	 mw-www-cli.log seems to come from maintenance/update.php
[09:05:55] <hashar>	 kostajh: I think that would be better to have $wgMWLoggerDefaultSpi to be set in includes/DevelopmentSettings.php
[09:06:09] <kostajh>	 hashar: ok, I am fine with that
[09:06:11] <hashar>	 which is where we set $wgDebugLogFile
[09:06:32] <hashar>	 I think that will be less surprising to have all the logging config at the same place
[09:06:44] <hashar>	 plus that saves us from having to release a new quibble + build images whenever we need to tune the settings
[09:06:55] <kostajh>	 I'll move it
[09:07:34] <wikibugs>	 (03Abandoned) 10Kosta Harlan: Logging: Use MWLoggerDefaultSpi [integration/quibble] - 10https://gerrit.wikimedia.org/r/763333 (https://phabricator.wikimedia.org/T225730) (owner: 10Kosta Harlan)
[09:08:02] <hashar>	 also the `mw-www-cli.log`  name is a bit confusing ;D
[09:11:07] <hashar>	 and totally unrelated
[09:11:32] <hashar>	 arturo proposed we spawn WMCS instances with a ton of memory and run the Quibble job entirely on tmpfs to save on disk I/O :D
[09:14:16] <wikibugs>	 (03Merged) 10jenkins-bot: phpbench: Use git-checkout instead of git-switch [integration/quibble] - 10https://gerrit.wikimedia.org/r/774379 (https://phabricator.wikimedia.org/T291549) (owner: 10Kosta Harlan)
[09:17:35] <wikibugs>	 (03CR) 10Kosta Harlan: "I84be19b7915163ff34ed31428c3c6864db43f698 for the core patch" [integration/quibble] - 10https://gerrit.wikimedia.org/r/763333 (https://phabricator.wikimedia.org/T225730) (owner: 10Kosta Harlan)
[09:17:53] <wikibugs>	 (03PS1) 10Hashar: release: Quibble 1.4.5 [integration/quibble] - 10https://gerrit.wikimedia.org/r/774410 (https://phabricator.wikimedia.org/T291549)
[09:18:06] <kostajh>	 re tmpfs, worth a shot i suppose
[09:18:35] <wikibugs>	 (03PS1) 10Hashar: release: Start 1.4.6 cycle [integration/quibble] - 10https://gerrit.wikimedia.org/r/774411
[09:20:23] <hashar>	 kostajh: ^ the painful release started :D
[09:21:30] <kostajh>	 :D
[09:21:37] <wikibugs>	 (03CR) 10Kosta Harlan: [C: 03+2] release: Start 1.4.6 cycle [integration/quibble] - 10https://gerrit.wikimedia.org/r/774411 (owner: 10Hashar)
[09:21:48] <wikibugs>	 (03CR) 10Kosta Harlan: [C: 03+2] release: Quibble 1.4.5 [integration/quibble] - 10https://gerrit.wikimedia.org/r/774410 (https://phabricator.wikimedia.org/T291549) (owner: 10Hashar)
[09:51:12] <wikibugs>	 (03Merged) 10jenkins-bot: release: Quibble 1.4.5 [integration/quibble] - 10https://gerrit.wikimedia.org/r/774410 (https://phabricator.wikimedia.org/T291549) (owner: 10Hashar)
[09:51:14] <wikibugs>	 (03Merged) 10jenkins-bot: release: Start 1.4.6 cycle [integration/quibble] - 10https://gerrit.wikimedia.org/r/774411 (owner: 10Hashar)
[10:47:08] <hashar>	 finally trying phpbench again https://integration.wikimedia.org/ci/job/mediawiki-phpbench-patch-docker/30/console :)
[10:51:23] <hashar>	 grmblblblbb
[10:51:31] <hashar>	 of course we fall back to memcached not being properly setup
[10:51:32] <hashar>	 :-\
[10:54:09] * hashar enables supervisord 
[10:54:10] <hashar>	 https://integration.wikimedia.org/ci/job/mediawiki-phpbench-patch-docker/31/console :]
[10:54:33] <hashar>	 00:01:23.687 INFO:quibble.commands:No phpbench entry found in composer.json
[10:54:33] <hashar>	 bah
[10:54:53] <hashar>	 I think cause the parent commit doesn't have a phpbench composer script
[10:59:04] <hashar>	 so I have send a dummy change https://gerrit.wikimedia.org/r/c/mediawiki/extensions/GrowthExperiments/+/774439/
[10:59:09] <hashar>	 and we got something at https://integration.wikimedia.org/ci/job/mediawiki-phpbench-patch-docker/32/console !
[11:19:40] <kostajh>	 hashar: nice, thank you!
[11:19:53] <kostajh>	 supervisord strikes again...