Page MenuHomePhabricator

MediaWiki\Extension\CentralAuth\Special\SpecialCentralAutoLogin::execute: Bad token: {loginResult}
Closed, ResolvedPublicPRODUCTION ERROR

Description

Error
  • mwversion: 1.46.0-wmf.24
  • timestamp: 2026-04-30T09:24:55.677Z
  • phpversion: 8.3.30
  • reqId: 54498841-4081-4345-8120-224647b66e94
  • Find reqId in Logstash
normalized_message
MediaWiki\Extension\CentralAuth\Special\SpecialCentralAutoLogin::execute: Bad token: {loginResult}
exception.trace
Impact
Notes

Steady increase over the last couple of days: https://logstash.wikimedia.org/goto/cccee76a68927819ae85533c64548f9

Screenshot 2026-04-30 at 10.29.02 AM.png (1,224×482 px, 49 KB)

Details

Request URL
https://es.wikipedia.org/wiki/Special:CentralAutoLogin/setCookies?type=*&useformat=*&usesul3=*

Event Timeline

It fixed itself:

image.png (2,029×465 px, 66 KB)

We also had an increase in "Lost session" errors on the dashboard in the same time frame: https://grafana.wikimedia.org/d/000000004/authentication-metrics?from=now-7d

image.png (815×455 px, 53 KB)
image.png (815×455 px, 52 KB)

The blue markers here highlight that it started with the "group1 to 1.46.0-wmf.26" deployment and ended with the "group2 to 1.46.0-wmf.26" deployment.

This was probably caused by the same cache inconsistency issue as T424935, which also resolved itself once all wikis were on the same MediaWiki version.

JTweed-WMF claimed this task.