[rabbitmq-discuss] Mirrored Queues Disappearing

[Plock, Justin (GE Corporate)]

Here's our scenario:

We run a 3 node RabbitMQ cluster (version 3.1.1) on 3 separate virtual machines.  All of our queues/exchanges are set to durable.  We have a policy called "HA" set up with the pattern of "^(?!amq\.).*" that as ha-mode: all and ha-sync-mode: automatic.  We have "cluster_partition_handling" set to "pause_minority".  Every time we experience a network partition, which seems to more often than it should, after forcefully killing RabbitMQ and restarting it, we also have to rebuild our missing queues and exchange bindings.  The virtual hosts, users and exchanges seems to be preserved, but we often lose a lot of our queues.

Is there anything we can do to prevent this?

Thanks.

-Justin
-------------- next part --------------
A non-text attachment was scrubbed...
Name: smime.p7s
Type: application/pkcs7-signature
Size: 3377 bytes
Desc: not available
URL: <http://lists.rabbitmq.com/pipermail/rabbitmq-discuss/attachments/20130611/ce2056fd/attachment.bin>