daniel added a subscriber: daniel.
daniel added a comment.

From looking at the code, it should actually be easy to avoid the fatal error 
without introducing a delay, as the proposed patch does. UpdateRepoOnMoveJob 
currently works like this:

1. verifyValid() calls normalizePageName(), and //silently// aborts the job if 
normalization fails (because the page isn't found under the new name for some 
reason).
2. applyChanges() calls normalizePageName() //again//, making another API call; 
if that call fails (e.g. because it's hitting another slave, which is more 
lagged), a //fatal// error is trigged, which lead to the log entry described 
above.

We could just fuse verifyValid with applyChanges, so we only need one API call 
to normalize the title; if that call fails, we would abort silently, as before. 
That would fix this bug, as reported: no more fatal errors in the log.

However, we should not silently skip updates just because of slave lag. We 
could just introduce a delay to reduce this issue, or have a way to force the 
API to query the master database.


TASK DETAIL
  https://phabricator.wikimedia.org/T76545

REPLY HANDLER ACTIONS
  Reply to comment or attach files, or !close, !claim, !unsubscribe or !assign 
<username>.

EMAIL PREFERENCES
  https://phabricator.wikimedia.org/settings/panel/emailpreferences/

To: daniel
Cc: Aklapper, aude, hoo, daniel, Wikidata-bugs



_______________________________________________
Wikidata-bugs mailing list
[email protected]
https://lists.wikimedia.org/mailman/listinfo/wikidata-bugs

Reply via email to