You are viewing a plain text version of this content. The canonical link for it is here.
Posted to issues@activemq.apache.org by "ASF subversion and git services (JIRA)" <ji...@apache.org> on 2016/03/15 15:53:33 UTC
[jira] [Commented] (AMQNET-521) LRUCache not working correctly,
which can cause Memory Leak and Slowdown
[ https://issues.apache.org/jira/browse/AMQNET-521?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15195414#comment-15195414 ]
ASF subversion and git services commented on AMQNET-521:
--------------------------------------------------------
Commit 1735085 from [~tabish121] in branch 'ActiveMQ/trunk'
[ https://svn.apache.org/r1735085 ]
Apply fix for problems with LRUCache not properly cleaning up after itself.
Fixes [AMQNET-AMQNET-521]. (See https://issues.apache.org/jira/browse/AMQNET-AMQNET-521)
> LRUCache not working correctly, which can cause Memory Leak and Slowdown
> ------------------------------------------------------------------------
>
> Key: AMQNET-521
> URL: https://issues.apache.org/jira/browse/AMQNET-521
> Project: ActiveMQ .Net
> Issue Type: Bug
> Components: ActiveMQ
> Affects Versions: 1.7.1
> Reporter: Markus Jung
> Assignee: Timothy Bish
> Fix For: 1.7.2, 1.8.0
>
> Attachments: LRUCache.cs.patch
>
>
> Reason: Duplicates are not handled correctly. Which causes one of the internal lists (entries) to grow over time until handling the numberous items increases processing overhead and cause a slowdown.
> Background: We are pulling a queue using n consumers with zero prefetch for maximum round robin. (Using Failover-Transport.)
> Processing a message can take time from split a second to several hours. Prefetching a message would not be good. Nether-the-less this is a good way to do it, we discovered increasing memory usage and slowing down of our "workers".
> After a worker processed a task a new call to consumer.Receive(timeout) causes to send a new MessagePull request, which is put into the messageCache using the same id.
> (ProcessMessagePull in ConnectionStateTracker)
> This causes the grow.
> Reason: In LRUCache.this method, the call to entries.Remove uses value and not currentValue.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)