Thompsonbry.systap added a comment. See http://trac.bigdata.com/ticket/1228 for some further thoughts on a root cause, some thoughts on how to create a stress test to replicate the problem, and some thoughts on how to fix this.
At this point I think we should focus on tests that interrupt the CreateKB or DestroyKB tasks (perhaps with other tasks to write on the existing namespaces per the tests referenced on that ticket). Those tests should be run both with group commit enabled and with it disabled. I suspect that the problem might only exist when group commit is NOT in use (and that it is due to the refactor to support group commit). The tests need to prompt abnormal failures in order to uncover incorrect abort handling logic. Once we can trigger the problem we can find ways to patch it in 1.5.1. I would like to change how this is handled in 1.5.2 in order to make such problems significantly less likely - effectively by building in some of the protections of the group commit code paths into the non-group commit code path. TASK DETAIL https://phabricator.wikimedia.org/T97468 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: Thompsonbry.systap Cc: JanZerebecki, Jdouglas, Beebs.systap, Thompsonbry.systap, Haasepeter, Manybubbles, Aklapper, Smalyshev, jkroll, Wikidata-bugs, aude, daniel _______________________________________________ Wikidata-bugs mailing list [email protected] https://lists.wikimedia.org/mailman/listinfo/wikidata-bugs
