Something like that would have drastically helped in catching T301433: Wikimedia\Rdbms\DBReadOnlyError: Database is read-only: The database is read-only until replication lag decreases. early.
See the list: https://logstash.wikimedia.org/goto/49e2ec8ced5be650e12c54e22dca6c92
| Status | Subtype | Assigned | Task | ||
|---|---|---|---|---|---|
| Open | None | T301742 Increase log level of rowsAffected > 1000 on database writes in transaction profiler to warning or error | |||
| Resolved | Umherirrender | T301641 GlobalBlocking purge expired must have a limit | |||
| Open | None | T303531 Write queries in ReadingListRepository::teardownForUser need a limit |
Unfortunately we need more clean ups before turning this on: https://logstash.wikimedia.org/goto/96f4a2de181b873b95b7a78bbe96151b
I will split it to per issue and create a subticket for each.
That's a bit different. The rowsAffected > 10,000 rows are for read queries but I'm talking about write queries here. Write queries are more important because they could cause read-only as the mentioned ticket did. I can create another ticket for 10K read rows for future clean up.
Change 766564 had a related patch set uploaded (by Ladsgroup; author: Amir Sarabadani):
[mediawiki/core@master] block: Add a limit to DatabaseBlockStore::purgeExpiredBlocks
Change 766564 merged by jenkins-bot:
[mediawiki/core@master] block: Add a limit to DatabaseBlockStore::purgeExpiredBlocks
Change 766850 had a related patch set uploaded (by Krinkle; author: Krinkle):
[mediawiki/core@master] block: Use $wgUpdateRowsPerQuery for purgeExpiredBlocks()
Change 766850 merged by jenkins-bot:
[mediawiki/core@master] block: Use $wgUpdateRowsPerQuery for purgeExpiredBlocks()
A couple more places and it'll be properly cleaned up https://logstash.wikimedia.org/goto/49e2ec8ced5be650e12c54e22dca6c92