activemq-users mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Tim Bain <>
Subject Re: Separate kahadb for queues and dead letter queues?
Date Fri, 29 Apr 2016 13:11:08 GMT
Using mKahaDB with a separate KahaDB instance for the DLQ is the best idea
I know of for dealing with the problem you describe.

A JIRA enhancement request exists for compacting KahaDB journals (  You
should vote for it, and if you felt like implementing a fix and submitting
a pull request, that would help make sure it gets implemented sooner rather
than later.
On Apr 29, 2016 6:39 AM, "Christian Schneider" <>

> We have the issue that our journals at a customer are very large. We
> analyzed the messages in the system and found that most of the time
> the normal queues stay at a very low number of messages near to 0. The
> only queues that hold messages for longer periods of time are the dead
> letter queues.
> The journals are much bigger than the dead letter queue contents.
> My assumption is that most of the journals only contain very few still
> needed messages and the ones they contain are the messages that now are in
> the dead letter queues.
> Still the journals can not be cleaned up of course.
> So what I would like to to is to use one kahadb for the regular queues and
> a different one for the dead letter queues. Does that make sense?
> I will try to do this using "Multi(m) kahaDB persistence adapter": See
> I also wonder if it would make sense that ActiveMQ cleans up its journals
> regularly. It could simply move all oldest messages to the newest jorunal
> to kind of pack if denser.
> Would that work?
> Christian
> --
> Christian Schneider
> Open Source Architect

  • Unnamed multipart/alternative (inline, None, 0 bytes)
View raw message