Last modified: 2014-08-28 20:23:35 UTC
Make sure that failures during indexing cause the job to be retried. They certainly ought to be retried but I'd just like to make sure one more time. Can reproduce by lowering the bulk threads to 1 and the queue to 0 and in vagrant and then running lots of job processes. We should get logged failures and retries.
Change 156837 had a related patch set uploaded by Manybubbles: Make sure that update errors are retried https://gerrit.wikimedia.org/r/156837
Change 156837 merged by jenkins-bot: Make sure that update errors are retried https://gerrit.wikimedia.org/r/156837