[ https://issues.apache.org/jira/browse/CASSANDRA-3295?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13120173#comment-13120173 ]
Radim Kolar commented on CASSANDRA-3295: ---------------------------------------- It crashes on 512m if you have to do compaction with about 20 tables. If you have to do major compaction on 400 tables in crashes even on 1G. Cassandra 1.0 is way better on memory use than 0.8 but still 512m is not enough for stable production operation. > reduce default heap size > ------------------------ > > Key: CASSANDRA-3295 > URL: https://issues.apache.org/jira/browse/CASSANDRA-3295 > Project: Cassandra > Issue Type: Improvement > Affects Versions: 1.0.0 > Reporter: Jonathan Ellis > Assignee: satish babu krishnamoorthy > Priority: Minor > Fix For: 1.0.0 > > Attachments: 3295.patch > > > With off-heap caching now the default and with auto-flushed memtables > demonstrated to do a good job with smaller heaps, I think it's time to reduce > heap sizes to decrease GC pause times. > How does this sound? > - half the ram, up to 512MB, or > - 1/4 the ram, up to 4 GB > i.e.: > max(min(1/2 ram, 512MB), min(1/4 ram, 4GB)) -- This message is automatically generated by JIRA. If you think it was sent incorrectly, please contact your JIRA administrators: https://issues.apache.org/jira/secure/ContactAdministrators!default.jspa For more information on JIRA, see: http://www.atlassian.com/software/jira