Page MenuHomePhabricator

[ops week] Troubleshoot and correct MediaWiki Wikitext history data after failures
Closed, ResolvedPublic

Description

The calculation of the MediaWiki Wikitext History dataset for the snapshot 2023-12 had some issues.

@fkaelin discovered that the size of the resulting datasets was off, with some wikis having a much bigger size than usual.
We soon identified this an another instance of the issue described in T342911.

This task is about the troubleshooting of this particular snapshot 2023-12,
and the correction of the data.

Acceptance Criteria

  • The data is in its production location as expected
  • The data is complete and not corrupted

Required

  • Identify what parts of the data are corrupted and why.
  • Find the quickest way to fix the data
  • Fix the data