Mpaa added a comment.
I would expect a failure rate around 0.5 % (2*5 / 1,727). If you increase step/len of page buffer, the failure will be more frequent.
TASK DETAIL
https://phabricator.wikimedia.org/T84944
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: Mpaa
Cc: pywikipedia-bugs, Aklapper, jayvdb, Mpaa
jayvdb added a comment.
That might be it, given random acts quite differently.
Also, as the failures are always (so far..) on test.wikidata, I am wondering if that might be a significant part of the problem. It has a low page count, from which to randomly select pages from. https://test.wikidata.org/wiki/Special:Statistics But not low enough that random should fail regularly, so maybe that isnt part of the problem.
TASK DETAIL
https://phabricator.wikimedia.org/T84944
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: jayvdb
Cc: pywikipedia-bugs, Aklapper, jayvdb, Mpaa
Mpaa added a subscriber: Mpaa.
Mpaa added a comment.
I guess each request in the QueryGenerator _Iter_ loop is independent from the previous. If one is unlucky enough, the random start of a batch might fail in an already seen interval?
TASK DETAIL
https://phabricator.wikimedia.org/T84944
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: Mpaa
Cc: pywikipedia-bugs, Aklapper, jayvdb, Mpaa