[
https://issues.apache.org/jira/browse/QPID-2004?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=12874509#action_12874509
]
Gordon Sim commented on QPID-2004:
----------------------------------
I'm certainly happy to have an alternative solution committed in place of this.
The clusters PollerDispatch uses the disconnected callback purely to detect
errors, as this occurs on shutdown now as well I added the ability to test
whether the Poller had been shutdown and avoided treating the disconnect as an
error in that case.
I should also note that the changes I committed only affect linux, so the issue
remains on windows.
> On shutting down the broker, the connection state of active connections is
> not freed
> ------------------------------------------------------------------------------------
>
> Key: QPID-2004
> URL: https://issues.apache.org/jira/browse/QPID-2004
> Project: Qpid
> Issue Type: Bug
> Components: C++ Broker
> Affects Versions: 0.5
> Environment: Linux
> Reporter: Gordon Sim
> Assignee: Gordon Sim
> Priority: Minor
>
> The IO layer (e.g. AsynchIOHandler for TCP) creates and holds a
> ConnectionCodec object by pointer for each connection. All connection state
> is rooted in that object. However for connections that are active when the
> broker is shutdown, those objects and all their internal state, is not freed
> but is leaked.
> As the broker is shutting down anyway, the leak is not nice but not critical
> (henc I've marked this as minor rather than major). However it can lead to
> e.g. Queue destructors not being called etc.
--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.
---------------------------------------------------------------------
Apache Qpid - AMQP Messaging Implementation
Project: http://qpid.apache.org
Use/Interact: mailto:[email protected]
