Wow, great turnaround time. Thanks!<br><br>> As a curiosity, how are you waiting to be sure that the messages are persisted?<br><br>I'm using transactions.<br><br>For what it's worth, this hang on broker start is happening much less than 10% of the time. It happened twice in 10 minutes one day, and the next day didn't happen at all.<br>
<br>For what it's worth, I now have Mnesia logging enabled so that if/when it does repro again, I'll have more details.<br><br>Matt<br><br><br><div class="gmail_quote">On Thu, Jun 21, 2012 at 10:53 AM, Francesco Mazzoli <span dir="ltr"><<a href="mailto:francesco@rabbitmq.com" target="_blank">francesco@rabbitmq.com</a>></span> wrote:<br>
<blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex">Hi Matt,<br>
At Thu, 21 Jun 2012 09:05:05 -0700,<br>
<div class="im">Matt Pietrek wrote:<br>
> Thanks Francesco.<br>
><br>
> Your script looks reasonably close to what I'm doing, except for a<br>
> couple of key differences:<br>
><br>
> All of your nodes are running on the same host. In my setup, each<br>
> node is in its own VM. I imagine this introduces additional<br>
> networking into the communication between nodes.<br>
<br>
</div>Yeah. I'm on my laptop these days, so if anything I'll a VM setup on<br>
Monday.<br>
<div class="im"><br>
> When you start Rabbit instances, you're doing it sequentially. In my<br>
> setup I start up in parallel on all three VMs via Capistrano. If you<br>
> can't go the multi-VM route, you may be able to somewhat simulate<br>
> this by starting the instances with an '&' at the end.<br>
<br>
</div>Well, I'm forking them one after the other and then waiting for them<br>
to actually start, so is not that sequential. I doubt that matters.<br>
<div class="im"><br>
> I use "killall -9 beam.smp", rather than just "killall beam.smp"<br>
<br>
</div>Right. This should not make much of a difference either. I tried with<br>
-9 and things were OK.<br>
<br>
As a curiosity, how are you waiting to be sure that the messages are<br>
persisted? I am using publisher confirms, see<br>
<a href="http://www.rabbitmq.com/extensions.html" target="_blank">http://www.rabbitmq.com/extensions.html</a> - that's why puka and not pika<br>
- but you need to make sure of that, otherwise you can get message<br>
loss or other strange behaviours.<br>
<br>
If you are trying to simulate what happens when the system crashes<br>
while sending messages, then it's different and of course you can't<br>
rely on any of the messages in being there when you restart them - but<br>
that does not justify the behaviour you're seeing anyway.<br>
<span class="HOEnZb"><font color="#888888"><br>
Francesco.<br>
</font></span></blockquote></div><br>