hoo added a comment.

I poked at potential DB related changes a bit and found none. The tables are consistent across the cluster AFAICT and the queries are being executed in the same way.

Also I quickly micro-benchmarked getConnection+reuseConnection which we do quite a lot in both DCs and couldn't find a difference big enough to explain this (while codfw seems to be like 30% faster, this is not nearly enough).

> $factory = MediaWiki\MediaWikiServices::getInstance()->getDBLoadBalancerFactory();
> $t0 = microtime( true );
> for ( $i = 0; $i < 500; $i++ ) { $repoLB = $factory->getMainLB( 'wikidatawiki' ); $db = $repoLB->getConnection( DB_REPLICA, [], 'wikidatawiki' ); $db->query( 'SELECT "1"' ); $repoLB->reuseConnection( $db ); }
> echo microtime( true ) - $t0;

Due to this I would suggest further poking at the interaction with Redis.


TASK DETAIL
https://phabricator.wikimedia.org/T205865

EMAIL PREFERENCES
https://phabricator.wikimedia.org/settings/panel/emailpreferences/

To: hoo
Cc: hoo, Joe, User-Marostegui, Banyek, jcrespo, Marostegui, Addshore, Aklapper, Nandana, jijiki, Lahi, Gq86, Darkminds3113, GoranSMilovanovic, QZanden, LawExplorer, Vali.matei, Volans, Volker_E, Wikidata-bugs, aude, GWicke, Dinoguy1000, Mbch331, Jay8g, Krenair, akosiaris
_______________________________________________
Wikidata-bugs mailing list
[email protected]
https://lists.wikimedia.org/mailman/listinfo/wikidata-bugs

Reply via email to