Last modified: 2014-01-27 21:58:47 UTC
Geowiki's private data repository shows no commit for 2013-12-15. Logs from geowiki computations at /a/geowiki/logs/20131115_20131215 on stat1 do not show a failure. /a/geowiki/data-private repo on stat1 is dirty. So it seems the data has been computed (but not committed). /a/geowiki/data-public repo on stat1 has a commit for 2013-12-15, but gerrit does not have it. Maybe just pushing the data-public commit to gerrit failed? (We had reports of connecting to gerrit fails sometimes.) That would explain what we're seeing, and the problem should disappear upon the next run of the cronjob.
Prioritization and scheduling of this bug is tracked on Mingle card https://wikimedia.mingle.thoughtworks.com/projects/analytics/cards/1324
Pushing worked again on the following day and failed again afterwards.
Change 103222 had a related patch set uploaded by QChris: Split committing and pushing limn files into separate steps https://gerrit.wikimedia.org/r/103222
Change 103222 merged by Milimetric: Split committing and pushing limn files into separate steps https://gerrit.wikimedia.org/r/103222
Hi Christian -- can you update the status of this bug? thanks, -Toby
(In reply to comment #5) > Hi Christian -- can you update the status of this bug? Not yet. The change got merged only yesterday (see comment #4). So today was the first day the new version got used. Since the bug's problem occurred only sporadic, I'd prefer to wait a few days and see if the problem does or does not surface again.
The problem did not surface again in the last two weeks. Let's assume our guess about the cause was correct, and the fix did what it was supposed to do.