[
https://issues.apache.org/jira/browse/CASSANDRA-5661?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13701497#comment-13701497
]
Pavel Yaskevich commented on CASSANDRA-5661:
--------------------------------------------
It's more of object pool, where each key has limited number of "equal" objects,
so if caller wants to read any portion of the file it would get an instance
which represents whole file, seek to appropriate position and do reading,
returning that instance to the pull when done. evicting block of buffers is
still required when files are compacted out which would be more frequent than
eviction by timer because use-cases usually don't drift in pattern. I also
raised question about what I think is a major problem here - max size of a file
being 5MB by default for CLS.
> 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: 1.2.7
>
> Attachments: CASSANDRA-5661.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
