You need to implement mirrored queues and clustering to provide high-availability for RabbitMQ: http://www.rabbitmq.com/ha.html You can see the implantation details in the following Puppet module: https://github.com/CiscoSystems/puppet-openstack-ha/tree/folsom_ha Regards, Daneyon Hansen On Mar 8, 2013, at 10:56 PM, "John Dewey" <john at dewey.ws> wrote: > I figure I will start here before asking elsewhere. > Has anyone successfully fronteded openstack's rabbitmq > layer with BigIPs? > > We are currently running folsom, and ran into the > following problem while performing failover tests. > The BigIP issues a "Connection reset by peer", when > a pool memeber's service is not available. After > a short delay the health checks will notice the failure > and take it out of service (as expected). > > A "Connection reset by peer" doesn't seem to be handled > by impl_kombu (and likely rightfully so). This results > in the message being lost. > > Any ideas? > > Thanks - > John > > _______________________________________________ > OpenStack-operators mailing list > OpenStack-operators at lists.openstack.org > http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-operators