<!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 3.2//EN">
<HTML>
<HEAD>
<META HTTP-EQUIV="Content-Type" CONTENT="text/html; charset=iso-8859-1">
<META NAME="Generator" CONTENT="MS Exchange Server version 6.5.7652.24">
<TITLE>RE: [rabbitmq-discuss] High cpu usage for Java RabbitMQ client?</TITLE>
</HEAD>
<BODY>
<!-- Converted from text/plain format -->
<P><FONT SIZE=2>Matthias,<BR>
<BR>
> You should not be using the tip. See<BR>
> <A HREF="http://lists.rabbitmq.com/pipermail/rabbitmq-discuss/2009-January/003164.html">http://lists.rabbitmq.com/pipermail/rabbitmq-discuss/2009-January/003164.html</A><BR>
<BR>
I used the directions from <A HREF="http://www.rabbitmq.com/build-server.html">http://www.rabbitmq.com/build-server.html</A> to download. If those are incorrect, then that needs to be fixed. I was referred to that page by an email a little bit ago to the list.<BR>
<BR>
>Looks fairly sane. The connection thread sits exactly where you'd expect<BR>
> it to be when not much is happening on the connection, namely in a<BR>
> socketRead0. There are a whole bunch of non-rabbit threads though,<BR>
> including UI threads. It would be good to eliminate those from our<BR>
> enquiries.<BR>
<BR>
Yes there are a lot of UI threads. The UI is entirely driven by messages from AMQP. <BR>
<BR>
What I did, in a different experiment, was disable posting to the AMQP broker. The UI is then doing nothing, and the system process time is the majority of effort. I'm thinking of running strace on this, to see what the specific issue is. My concern is that, for some reason, the call is not blocking in a nice way and is causing excess looping somewhere.<BR>
<BR>
I will develop a non-UI process that does the same things and test it. If I can reproduce this result then I can post that code.<BR>
<BR>
Thanks for your help,<BR>
Rob<BR>
</FONT>
</P>
</BODY>
</HTML>