Hi,

Sounds like you reached the client or server timeout.
So either you increase any values (or both) or you implement some kind
of keepalive method inside your application in order to avoid HAProxy
to shut it down.

The logs may provide some usefull information about which time out has expired.

cheers


On Fri, Feb 10, 2012 at 4:20 PM, Robert Recchia <[email protected]> wrote:
> We are trying to use haproxy to load balance rabbitmq nodes in a cluster.
>
> Here is the config we are using
>
>
> listen rabbitmq1-cluster-5672 192.168.72.26:5672
>        balance roundrobin
>        server atuapp22 192.168.72.71:5672 check port 5672
>        server atuapp18 192.168.72.67:5672 check port 5672
>
>
>
> These are the global settings
>
>
>
>
> # Global settings
> #---------------------------------------------------------------------
> global
>    log         127.0.0.1 local2
>    #chroot      /var/lib/haproxy
>    pidfile     /var/run/haproxy.pid
>    maxconn     4000
>    user        haproxy
>    group       haproxy
>    daemon
>    # this line is for hatop
>   #  stats socket /var/run/haproxy/haproxy.sock user haproxy group
> haproxy mode 0666
>
>
>
>
>    stats socket /var/lib/haproxy/stats
>
>
>
>
> #---------------------------------------------------------------------
> # common defaults that all the 'listen' and 'backend' sections will
> # use if not designated in their block
> #---------------------------------------------------------------------
> defaults
>    mode        tcp
>    log         global
>    option      dontlognull
>    maxconn     60000
>    retries     3
>    option      srvtcpka
>    option      clitcpka
>    timeout connect 5000ms
>    timeout client 50000ms
>    timeout server 50000ms
>
>
>
>
>
> The connection seems to always always close after 50 seconds and the
> rabbitmq Java client gets a EOF connection.  How can we avoid the TCP
> connection closing after 50 seconds and stay open permanently
> --
> Robert Recchia
>

Reply via email to