milleruntime commented on pull request #1899: URL: https://github.com/apache/accumulo/pull/1899#issuecomment-771930792
I tested this enhancement on Uno with 2 tservers running 2 RW MultiTable jobs and a Conditional job for increased pressure. It worked well and I saw a few situations where the largest tablet to be flushed was being deleted and dropped from the memory report. I was able to get the cluster up to 200+ tables with 2000+ tablets before the GC started running low on memory. <pre> 2021-02-02T13:59:35,144 [tablet.Tablet] DEBUG: Table 6z is being deleted so don't flush 6z;4;3 2021-02-02T13:59:35,144 [tserver.TabletServerResourceManager] DEBUG: Cleaned up report for closed/deleted tablet 6z;4;3 2021-02-02T13:59:35,144 [tserver.TabletServerResourceManager] DEBUG: Ignoring memory manager recommendation: not minor compacting closed tablet 6z;4;3 </pre> ---------------------------------------------------------------- This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. For queries about this service, please contact Infrastructure at: [email protected]
