[
https://issues.apache.org/jira/browse/QPID-7150?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15204380#comment-15204380
]
Alan Conway commented on QPID-7150:
-----------------------------------
That narrows it down a lot. The primary does some tracking of queue state even
when no backup is connected, in order to be ready when a backup connects. That
is probably where the problem lies.
> HA memory leak in primary broker when overwriting messages in a ring queue
> --------------------------------------------------------------------------
>
> Key: QPID-7150
> URL: https://issues.apache.org/jira/browse/QPID-7150
> Project: Qpid
> Issue Type: Bug
> Components: C++ Clustering
> Affects Versions: 0.32
> Reporter: Alan Conway
> Assignee: Alan Conway
> Fix For: qpid-cpp-next
>
>
> From https://bugzilla.redhat.com/show_bug.cgi?id=1318180
> HA memory leak in primary broker when overwriting messages in a ring queue
> ReplicatingSubscription accumulates IDs of dequeued messages to send on
> dispatch. It should clear the accumulated IDs once sent. Due to a merge error,
> since:
> 014f0f3 QPID-4327: HA TX transactions: basic replication.
> The ID set is not cleared, causing it to accumulate memory slowly.
> This leak would be particularly noticeable on a busy ring-queue since a
> ring-queue generates a dequeue event for every enqueue once it reaches its max
> size.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
