activemq-users mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Kevin Burton <bur...@spinn3r.com>
Subject Re: ActiveMQ performance during bulk queue GC.
Date Sun, 22 Feb 2015 04:54:08 GMT
Also, I realized there’s another bug here.  A write lock is acquired even
in the situation where no queues actually need to be GCd.  It’s not massive
in the grand scheme of things but it probably ends up with brief moments
where the broker isn’t serving any messages.

On Sat, Feb 21, 2015 at 8:42 PM, Kevin Burton <burton@spinn3r.com> wrote:

> I’m still tracking down this bulk queue GC bug I’ve been dealing with.
>
> Right now I’m working on a test to duplicate the problem.
>
> However, in the mean time I noticed this:
>
> http://pastebin.com/PwPJQLNu
>
> … in regionBroker.
>
> Essentially, the code holds the write lock during the entire time of the
> GC for *all* the queues.
>
> In my experience, on disk, it takes 100ms to GC a queue (at least under
> load)
>
> It seems like a better strategy here would be to have the GC interval be
> very low, say 5 seconds, but then have a getMaxPurgedDestinationsPerSweep
> value low.. say 1-5 …
>
> The problem is that there is no way to change this value anywhere.  There
> aren’t any configuration directives for this.  From the source, it looks
> like it’s only used within testing.
>
> Perhaps this should become a config directive?
>
> Kevin
>
> --
>
> Founder/CEO Spinn3r.com
> Location: *San Francisco, CA*
> blog: http://burtonator.wordpress.com
> … or check out my Google+ profile
> <https://plus.google.com/102718274791889610666/posts>
> <http://spinn3r.com>
>
>


-- 

Founder/CEO Spinn3r.com
Location: *San Francisco, CA*
blog: http://burtonator.wordpress.com
… or check out my Google+ profile
<https://plus.google.com/102718274791889610666/posts>
<http://spinn3r.com>

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