[jboss-jira] [JBoss JIRA] (AS7-6254) Memory leak caused by retained connection ids in RemotingConnectorServer superclass

[Taras Tielkes (JIRA)]

    [ https://issues.jboss.org/browse/AS7-6254?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=12743089#comment-12743089 ] 

Taras Tielkes commented on AS7-6254:
------------------------------------

Note that this issue is still present in 7.1.3 EAP.

Steps to reproduce:
1) Install clean 7.1.3 JBoss instance
2) Add JMX user using the provided admin script
3) Run a client that repeatedly creates a JMX connection (using the {{service:jmx:remoting-jmx}} protocol, using the provided {{jboss-client.jar}} library)

After sufficient time, depending on the configured heap size, you'll see the JBoss server JVM run out of memory.

I have an {{hprof}} heap dump that shows the leak reproduced by the steps above. If you're interested in looking at that, let me know where to upload it.
                
> Memory leak caused by retained connection ids in RemotingConnectorServer superclass 
> ------------------------------------------------------------------------------------
>
>                 Key: AS7-6254
>                 URL: https://issues.jboss.org/browse/AS7-6254
>             Project: Application Server 7
>          Issue Type: Bug
>          Components: JMX
>    Affects Versions: 7.1.1.Final
>            Reporter: Taras Tielkes
>            Assignee: Stuart Douglas
>
> We're running 7.1.1, with a patch applied for REMJMX-45 to limit the worst leaks coming from the JMX subsystem.
> However, even with this patch applied we can only survive for a few days in a production-like scenario.
> It seems that {{org.jboss.remotingjmx.RemotingConnectorServer}} never calls the {{connectionClosed()}}/{{connectionFailed()}} methods in its superclass.
> As such, the connection ids that are stored in the field {{javax.management.remote.JMXConnectorServer#connectionIds}} are never released.
> Given sufficient connections made to a running instance of 7.1.1 (for example, various monitoring tools), an out-of-memory end-state is inevitable.

--
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