You are viewing a plain text version of this content. The canonical link for it is here.
Posted to commits@cassandra.apache.org by "Héctor Izquierdo (JIRA)" <ji...@apache.org> on 2011/07/09 15:27:16 UTC
[jira] [Created] (CASSANDRA-2875) Increase index_interval and
reopen sstables on low heap situations
Increase index_interval and reopen sstables on low heap situations
------------------------------------------------------------------
Key: CASSANDRA-2875
URL: https://issues.apache.org/jira/browse/CASSANDRA-2875
Project: Cassandra
Issue Type: Improvement
Components: Core
Affects Versions: 0.8.1
Reporter: Héctor Izquierdo
One of the reasons that can cause an OOM is key indexes. Of course you can tune it, but that's after your node has crashed. Events like repair can cause a much bigger memory pressure than expected on normal operation. As part of the measures taken when heap is almost full it would be good if key indexes could be shrank. I don't know how indexes are stored in memory but I guess it would be possible to remove entries without rereading all sstables.
--
This message is automatically generated by JIRA.
For more information on JIRA, see: http://www.atlassian.com/software/jira
[jira] [Resolved] (CASSANDRA-2875) Increase index_interval and
reopen sstables on low heap situations
Posted by "Jonathan Ellis (JIRA)" <ji...@apache.org>.
[ https://issues.apache.org/jira/browse/CASSANDRA-2875?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]
Jonathan Ellis resolved CASSANDRA-2875.
---------------------------------------
Resolution: Won't Fix
Sorry, it's too hard to tell when "low heap" is a transitory condition vs when the JVM is really in trouble. We do our best by cutting row cache and flushing memtables (both relatively low-impact) but reopening sstables would cause more damage than good in a lot of situations.
> Increase index_interval and reopen sstables on low heap situations
> ------------------------------------------------------------------
>
> Key: CASSANDRA-2875
> URL: https://issues.apache.org/jira/browse/CASSANDRA-2875
> Project: Cassandra
> Issue Type: Improvement
> Components: Core
> Affects Versions: 0.8.1
> Reporter: Héctor Izquierdo
>
> One of the reasons that can cause an OOM is key indexes. Of course you can tune it, but that's after your node has crashed. Events like repair can cause a much bigger memory pressure than expected on normal operation. As part of the measures taken when heap is almost full it would be good if key indexes could be shrank. I don't know how indexes are stored in memory but I guess it would be possible to remove entries without rereading all sstables.
--
This message is automatically generated by JIRA.
For more information on JIRA, see: http://www.atlassian.com/software/jira