Resolve "Tape server querying DB at very high rate"
Summary
This MR is to fix this specific cause of high DB queries: #155 (comment 6266313)
Requires manual tests in pre-production
NO. The high query rate is only evidenced when there are many tape servers and a high rate of retrieve requests. Best way to test this is in production:
- Get an updated query rate from Nilo
- Deploy the version of CTA containing this fix
- Get a new query rate and it should hopefully be much lower
References
Closes #155 (closed)
Merge request reports
Activity
added File catalogue Scheduler Tape server workflowIn Progress labels
requested review from @dhsmith
assigned to @mdavis
mentioned in issue #155 (closed)
added 1 commit
- 0c2bd45a - Also purge tape cache when purging retrieve stats cache
added 1 commit
- 28ac7968 - Handles corner case where second candidate vid gets evicted from the cache
added 5 commits
Toggle commit listHi Michael. Looks fine. In updateRetrieveQueueStatisticsCache I wondered about also faking the entry in g_tapeStatuses if one didn't exist, but as far as I can see usually the entries should exist, probably only a corner case of just-expired entry where it may not, in which case adding the entry g_tapeStatues in the usual way in selectBestRetrieveQueue should be a small number of queries.
That's all I can think of, I'll add the 'approve' mark to this MR.
enabled an automatic merge when the pipeline for be158c83 succeeds
added 5 commits
Toggle commit listenabled an automatic merge when the pipeline for a738cce4 succeeds