[18:32:47] <TK-999_>	 good evening, it seems on SMW 2.4, it is possible under certain conditions for a mass delete of pages to trigger a huge message queue spike
[18:33:01] <TK-999_>	 http://narutofanon.wikia.com/index.php?title=Special%3ALog&type=delete&user=SahaTo&limit=263 - 263 pages deleted in 15 minutes generated over 1mln background tasks
[18:35:04] <TK-999_>	 interestingly the deletes themselves (which finished at 19:04 CEST) generated a backlog of "only" 30K tasks, then at around 19:23 CEST something happened which caused the job queue to rapidly spike to 1mln
[18:36:03] <TK-999_>	 do you have any pointers on where to continue the investigation?
[19:17:29] <TK-999_>	 some more details: in previous investigations we tied the occurrence to this to the fact we use a parallel execution task queue as opposed to core MW's sequential one - for this reason SMW queue has specifically lower concurrency
[19:17:46] <TK-999_>	 however this can trigger such issues sometimes (usually every few months)