<div dir="ltr">Thank you for reply,<div><br></div><div style>Unfortunately our client library does not support heartbeats so it is being disabled on connect.</div><div style>But problem disappeared after we upgraded RabbitMQ to 3.0.4 AND changed system tcp_keepalive_time to reasonable small value.</div>
<div style><br></div><div style>Now we have</div><div style><div>$ cat /proc/sys/net/ipv4/tcp_keepalive_time</div><div>14</div><div>$ cat /proc/sys/net/ipv4/tcp_keepalive_intvl </div><div>14</div><div>$ cat /proc/sys/net/ipv4/tcp_keepalive_probes </div>
<div>4</div><div style> and everything is fine with connection counts.</div><div style><br></div><div style>Thank you</div></div></div><div class="gmail_extra"><br><br><div class="gmail_quote">On 2 April 2013 13:56, 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 29/03/13 08:45, Dmitry Saprykin wrote:<br>
> When everything goes bad (rabbitmqctl list_connections | wc -l) twice<br>
> as much connections than (netstat -nt | wc -l)<br>
<br>
</div>Does the logfile contain anything beyond the handshake time-out message?<br>
Are there any entries from the time when the counts start to diverge?<br>
<br>
What state are the connections in when the failure occurs? Can you show<br>
some representative output from "rabbitmqctl list_connections" during<br>
the error?<br>
<br>
Have you tried using AMQP heartbeats instead of TCP keepalives?<br>
<span class="HOEnZb"><font color="#888888"><br>
<br>
<br>
<br>
-Emile<br>
<br>
<br>
<br>
<br>
<br>
</font></span></blockquote></div><br></div>