Page MenuHomePhabricator

Enable reindexing the Commons "File" index in Cloudelastic by default
Closed, ResolvedPublic2 Estimated Story Points

Description

User story: As a user of cloudelastic, I want the Commons File index to be up to date so I have access to accurate & timely data.

Commons has the usual indexes—content, general, and archive—plus a file index, which other wikis do not have. The reindexing process handles them in a fixed order: content, general, archive, file.

The Cloudelastic environment is not private, so the archive index is not created. It appears that this is reported as an "error", which stops the reindexing process, so no attempt is made to update the file index.

  1. A somewhat kludgy approach would be to move the logic that adds the file index to the list for Commons ahead of the archive index.
  2. A better approch would be to allow the archive index to "fail successfully" on Cloudelastic, so the file index can continue.
  3. Another alternative might be to be smarter about knowing what indexes are allowed where and not trying to index archive on Cloudelastic, but that could be brittle in the long term.

Acceptance Criteria: Reindexing Commons on the Cloudelastic cluster at least makes an attempt to update the File index (it may fail for other reasons, as they do).

Event Timeline

TJones renamed this task from Enable reindexing the Commons "File" index in Cloudelastic to Enable reindexing the Commons "File" index in Cloudelastic by default.Apr 14 2021, 7:40 PM
MPhamWMF moved this task from needs triage to elastic / cirrus on the Discovery-Search board.

I'm running into this again, so I'm going to try to go ahead and fix it. (Using Approach #2)

TJones set the point value for this task to 2.
TJones moved this task from Incoming to In Progress on the Discovery-Search (Current work) board.

Change 699962 had a related patch set uploaded (by Tjones; author: Tjones):

[mediawiki/extensions/CirrusSearch@master] Change error on reindexing archive on non-private cluster to a warning

https://gerrit.wikimedia.org/r/699962

Change 699962 merged by jenkins-bot:

[mediawiki/extensions/CirrusSearch@master] Change error on reindexing archive on non-private cluster to a warning

https://gerrit.wikimedia.org/r/699962