Wikis were unreachable for a brief period of time of around 12 minutes in total, split in 2 spikes: 08:24-08:30 and 08:39-08:45 UTC

The root cause seemed to be a network slowdown caused by ongoing maintenance on db1099- which caused fallout on all of s8 section, with later expanded to other sections too (s1 was also hosted on db1099, plus other wikis require connecting to wikidata database (s8).

More specifically it seems to have been caused by db1099 (an s8 and s1 replica) that was rebooted for maintenance and slowly repooled while at the same time there was a network transfer from the same host. The host became slow to respond, but not enough to be considered down and depooled automatically. This caused a cascade effect on DBs on the same section (S8) and because S8 is read by every wiki page, causing a cascade effect on all wikis, manifesting in the page for too many PHP-FPM workers busy.

This mostly affected logged-in users (editors) as cache layer was for the most part unaffected (except misses).

https://wikitech.wikimedia.org/wiki/Incident_documentation/2022-03-10_MediaWiki_availability