Open Stack

Hi everyone,

I see advice in replacing RabbitMQ with ZeroMQ. I've been running 2
clusters Liberty/CentOS7 with RabbitMQ now for while. The larger
cluster consists of 3x Controllers and 4x Compute nodes. RabbitMQ is
running is HA mode as per:
http://docs.openstack.org/ha-guide/shared-messaging.html#configure-rabbitmq-for-ha-queues.

I have several times experienced issue propagating through OpenStack
Networking, Compute, Storage services boiling down to errors in
RabbitMQ service. For instance, these seem to have happened after
network or DC interrupts, the Rabbit service as not in a robust or
reliable manner resumed working.

Here are some error examples:

=ERROR REPORT==== 7-Aug-2016::08:01:58 ===
Channel error on connection <0.6967.0> (192.168.2.14:34370 ->
192.168.2.12:5672, vhost: '/', user: 'openstack'), channel 1:
operation basic.consume caused a channel exception not_found: "failed
to perform operation on queue 'conductor_fanout_c0c8929ac9' in vhost
'/' due to timeout"

=ERROR REPORT==== 10-Aug-2016::12:43:31 ===
closing AMQP connection <0.2363.0> (192.168.2.12:55846 -> 192.168.2.12:5672):
missed heartbeats from client, timeout: 60s

etc..

I have thought of replacing RabbitMQ with ZeroMQ if that were to bring
a more resilient MQ service for my OpenStack cluster.  While reading
about Rabbit vs. ZeroMQ I have found they've each got their strengths,
RabbitMQ: good for Queue integrity, little slower, and Zero: simple,
optimized for network transfer, no broker, de-centralized by design.


I would highly appreciate if anyone running ZeroMQ can advice on the
experience in Production, or any issues experienced? Also if there are
any links or resources on how to go about replacing RabbitMQ with Zero
would be much appreciated.

thx,will