[ https://issues.apache.org/jira/browse/CASSANDRA-5661?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13707992#comment-13707992 ]
Pavel Yaskevich commented on CASSANDRA-5661: -------------------------------------------- As default queue size is set to 512MB the tests I did weren't actually using expiry on replacement (as there were no compaction nor dataset was been enough) so I shouldn't have touched the finalization path. I re-run with updated MultiwayPool code and saw the same results as previously stated in the steady state (which is expected), but when I add compaction in the mix (by overwriting data), latencies become shaky with MultiwayPool most notably on 99.9th percentile where random spikes by 10-15 ms are observed, FileCacheService 99.9th stays almost constant. > Discard pooled readers for cold data > ------------------------------------ > > Key: CASSANDRA-5661 > URL: https://issues.apache.org/jira/browse/CASSANDRA-5661 > Project: Cassandra > Issue Type: Bug > Components: Core > Affects Versions: 1.2.1 > Reporter: Jonathan Ellis > Assignee: Pavel Yaskevich > Fix For: 2.0 > > Attachments: CASSANDRA-5661-multiway-per-sstable.patch, > CASSANDRA-5661.patch, CASSANDRA-5661-v2-global-multiway-per-sstable.patch, > DominatorTree.png, Histogram.png > > > Reader pooling was introduced in CASSANDRA-4942 but pooled > RandomAccessReaders are never cleaned up until the SSTableReader is closed. > So memory use is "the worst case simultaneous RAR we had open for this file, > forever." > We should introduce a global limit on how much memory to use for RAR, and > evict old ones. -- This message is automatically generated by JIRA. If you think it was sent incorrectly, please contact your JIRA administrators For more information on JIRA, see: http://www.atlassian.com/software/jira