<div dir="ltr">I changed the keepalived tcp_check by other type called misc_check ( a check that can be done by yourself )<div>This check close the connection ( a simple nc execution ) , and everything start to work again. And the memory stop growing and got stabilized.</div>
</div><div class="gmail_extra"><br><br><div class="gmail_quote">On Tue, Oct 1, 2013 at 11:01 AM, Emile Joubert <span dir="ltr"><<a href="mailto:emile@rabbitmq.com" target="_blank">emile@rabbitmq.com</a>></span> wrote:<br>
<blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex"><br>
Hi,<br>
<div class="im"><br>
On 01/10/13 08:56, Santi Nuņez-Cacho wrote:<br>
> I disable TCP_CHECK on keepalived configuration and the memory stabilized.<br>
> @emile, do you use a specific configuration of keepalived that would<br>
> bypass this problem?<br>
<br>
</div>Thanks for the network packet capture. This showed that connections are<br>
being forced closed using TCP resets. I've managed to reproduce the<br>
issue and a fix will follow shortly. STOMP is the only protocol<br>
supported by RabbitMQ affected by this issue.<br>
<br>
A workaround in the meantime would be to use a configurations that does<br>
not close the network connections by resetting.<br>
<span class="HOEnZb"><font color="#888888"><br>
<br>
<br>
-Emile<br>
<br>
<br>
<br>
<br>
</font></span></blockquote></div><br></div>