[rabbitmq-discuss] Missing Durable Queues on Startup

Chris Larsen clarsen at llnw.com
Tue May 8 16:12:40 BST 2012

Thanks Steve!

> a) are all the nodes in your cluster disk nodes?

Yep, everything is a disk node

> b) was the node that you stop/started a disk node or a ram node?

The node stopped/started was a disk node

> c) when the cluster was restarted did you start all the disk nodes

We didn't restart the whole cluster, just these two nodes in the cluster.
The other nodes were running happily.

> When you say:
>> We're not using HA in rabbit itself, the queues are just persistent
>> and durable on each node in the cluster.

> Can you be more precise? I assumed that the node you restarted was the
one that held the queues you were missing.

Yeah, I read about the HA mirrored queues and such but due to our
throughput and the way rabbit replicated at the time, it made more sense
for us to use DRBD. So we aren't using any of Rabbit's HA features at this

> You appear to be running version 2.5.1. Clustering has had some fixes
applied to it since then.
> I would recommend upgrading to the latest release.

I'll look through the change logs and maybe upgrade our dev cluster and
see how it goes. Thanks!

More information about the rabbitmq-discuss mailing list