rmuir commented on issue #14999: URL: https://github.com/apache/lucene/issues/14999#issuecomment-3130328346
> Under such conditions, lock wait times can exceed 100ms, depending on the size of the parentReaders set. I have trouble believing this (think about the reported set size which isn't big and how slow you are claiming it to be). I'm not sure what profiler was used, but they frequently struggle with this stuff and report "ghosts". What was the dataset size used to measure p99? I'm asking because of the typical bug at amazon where folks try to measure such high P values but without applying the statistical rigor to do it correctly. If this map was truly the bottleneck from all the entries coming and going, I'd expect to see stuff such as `java.util.WeakHashmap.expungeStaleEntries()` in your profile, which leads me to believe it is a profiler ghost. -- 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. To unsubscribe, e-mail: issues-unsubscr...@lucene.apache.org For queries about this service, please contact Infrastructure at: us...@infra.apache.org --------------------------------------------------------------------- To unsubscribe, e-mail: issues-unsubscr...@lucene.apache.org For additional commands, e-mail: issues-h...@lucene.apache.org
