[
https://issues.apache.org/jira/browse/CASSANDRA-5661?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13701637#comment-13701637
]
Pavel Yaskevich commented on CASSANDRA-5661:
--------------------------------------------
You got me wrong, I'm not trying to debate anything with you and minute details
are proven to be the hardest sometimes... I'm just trying so say, to Jonathan
most of all, if we go with borrowing approach and put cache in front of each
row read we would suffer additional latency on every read (even if it's 1-3 ms,
that adds to every file read on each request) of the row where people already
report order of magnitude worse latencies on >= p99, am I right [~tjake]?
P.S. am I sorry I forgot to bow before I darred to speak up to such master who
solved all your problems and who's modesty doesn't know it's limits...
> 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
