[rabbitmq-discuss] If 2 nodes out a 3 node cluster, the third one becomes unresponsive until one of the nodes is brought back.

Matthias Radestock matthias at rabbitmq.com
Wed Aug 7 10:54:18 BST 2013

On 06/08/13 16:47, Yamil Einar Asusta Santos wrote:
> I have been testing my cluster and I have come across an unexpected
> behavior. As explained in the subject, the cluster runs well on 3 nodes.
> If I bring 1 node down, it still runs smoothly. But if I bring a second
> node down, the third one becomes unresponsive.
> This is what the third node provides after running "rabbitmqctl report":
>     Reporting server status on {{2013,8,6},{15,19,8}}
>       ...
>     Error: {aborted,{no_exists,rabbit_vhost}}
> Then if I bring 1 or 2 of my nodes back up, the third one becomes
> responsive and everything is back to normal.
> Here is the report after bringing the nodes up again:
> [...]
>     Cluster status of node rabbit at qcluster1 ...
>     [{nodes,[{disc,[rabbit at qcluster2,rabbit at qcluster0]},{ram,[rabbit at qcluster1]}]},

You have two disk and one ram node. If you stop both disk nodes, the 
remaining ram node will not be happy.


More information about the rabbitmq-discuss mailing list