milleruntime commented on pull request #1887: URL: https://github.com/apache/accumulo/pull/1887#issuecomment-770998940
I was doing some testing using Uno running 2 RW MultiTable jobs and saw a bad situation with this change. The `LargestFirstMemoryManager` will select the largest tablets to be flushed (default for _tserver.compaction.minor.concurrent.max_ is 4) and if they all happen to be in the table being deleted, no tablets will get flushed and things on the tserver will get stuck. Something is preventing the tablets from being unloaded and this new check stops them from being flushed. I am not sure why they still aren't being unloaded but we may need to revert this change. ---------------------------------------------------------------- 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]
