You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@qpid.apache.org by "Rob Godfrey (JIRA)" <ji...@apache.org> on 2015/07/06 13:07:05 UTC

[jira] [Created] (QPID-6621) [Java Broker] Slow performance of Shared Groups message grouping on large queues

Rob Godfrey created QPID-6621:
---------------------------------

             Summary: [Java Broker] Slow performance of Shared Groups message grouping on large queues
                 Key: QPID-6621
                 URL: https://issues.apache.org/jira/browse/QPID-6621
             Project: Qpid
          Issue Type: Bug
          Components: Java Broker
    Affects Versions: 0.32, 0.30, 0.28
            Reporter: Rob Godfrey
            Assignee: Rob Godfrey
             Fix For: 6.0 [Java]


In shared group mode, when a consumer gets to the point where it has no more messages from a group acquired by it, the consumer effectively relinquishes control of the group.  At this point all messages from the group become available to other consumers... which means that if they have gone "past" any such messages then then need to be reset.

Currently this is achieved by iterating over the group and finding the next available member of the group, and then making sure the consumers are set to before that point.

However if the queue is large, and the common case is that there are no members of the group, then every consumption of a message leads to iterating over the whole queue.

It is preferable to keep a track of which messages consumers may have skipped over, and simply to reset them to the least of these values.



--
This message was sent by Atlassian JIRA
(v6.3.4#6332)

---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@qpid.apache.org
For additional commands, e-mail: dev-help@qpid.apache.org