Incidents/2023-06-06 dawiktionary/dawiki/svwiktionary partial outage

From Wikitech

document status: draft


Incident metadata (see Incident Scorecard)
Incident ID dawiktionary/dawiki/svwiktionary partial outage Start 2023-06-06 00:30:00
Task T338193 End 2023-06-06 03:49:00
People paged 14 Responder count 4
Coordinators cwhite Affected metrics/SLOs
Impact All users of dawiktionary, dawiki, and svwiktionary sometimes received fatal exceptions while browsing.

All users of dawiktionary, dawiki, and svwiktionary sometimes received fatal exceptions while browsing. Logs indicated MediaWiki was encountering a data issue which also entered the object cache. The source of the problem was traced back to a data migration implementation bug in how bash interprets double-quotes (") characters.


All times in UTC.

  • 2023-05-31 15:45 low level of 500 errors on dawiktionary - data encoding migration begins: task T128155
  • 2023-06-06 00:30 errors dramatically increase - OUTAGE BEGINS
  • 2023-06-06 01:35 UBN task T338193 filed
  • 2023-06-06 02:36 Problem escalated to SRE via Klaxon - investigation begins
  • 2023-06-06 03:36 Investigation yields possible data issue - Escalated to DBA running the encoding migration
  • 2023-06-06 03:49 Responding DBA detects and implements the fix - Client-facing errors stop - OUTAGE ENDS


The issue was noticed by a volunteer and reported after business hours via IRC. Another volunteer saw the IRC message and escalated to SRE via Klaxon.


What went well?

  • Klaxon notified SRE and a number of engineers responded quickly.
  • VO made it easy to escalate the issue to another engineer.
  • Once a DBA arrived, the issue was fixed quickly and the root cause identified.
  • We intentionally started with small wikis on legacy encoding. Dutch Wikipedia and English Wikipedia could have had the outage instead.

What went poorly?

  • The problem surfaced inconsistently when running spot-checks.
  • The problem resolution required deep technical knowledge of MediaWiki data formats and internal caching systems.

Where did we get lucky?

  • It was early in the morning for them, but a DBA with the necessary expertise responded to the call when requested.

Links to relevant documentation

  • N/A


  • TODO


Incident Engagement ScoreCard
Question Answer


People Were the people responding to this incident sufficiently different than the previous five incidents? yes
Were the people who responded prepared enough to respond effectively yes
Were fewer than five people paged? no
Were pages routed to the correct sub-team(s)? yes
Were pages routed to online (business hours) engineers?  Answer “no” if engineers were paged after business hours. no
Process Was the "Incident status" section atop the Google Doc kept up-to-date during the incident? no Not Used
Was a public entry created? no
Is there a phabricator task for the incident? yes
Are the documented action items assigned? yes
Is this incident sufficiently different from earlier incidents so as not to be a repeat occurrence? yes
Tooling To the best of your knowledge was the open task queue free of any tasks that would have prevented this incident? Answer “no” if there are

open tasks that would prevent this incident or make mitigation easier if implemented.

Were the people responding able to communicate effectively during the incident with the existing tooling? yes
Did existing monitoring notify the initial responders? no Klaxon
Were the engineering tools that were to be used during the incident, available and in service? yes
Were the steps taken to mitigate guided by an existing runbook? no
Total score (count of all “yes” answers above)