[rabbitmq-discuss] Clustering question

Dmitriy Samovskiy dmitriy.samovskiy at cohesiveft.com
Tue Sep 16 21:34:56 BST 2008


Hi all,

I am running a RabbitMQ cluster - rabbit at host1 and rabbit at host2. After a loss of network 
connectivity between host1 and host2, each node fails to note its counterpart's presence 
even after connectivity is restored.

rabbitmqctl status on host1 says:
{nodes,[rabbit at host1,rabbit at host2]},
{running_nodes,[rabbit at host1]}

rabbitmqctl on host2 says:
{nodes,[rabbit at host1,rabbit at host2]},
{running_nodes,[rabbit at host2]}

This does not change if I start a producer and start publishing, or if I start a consumer. 
I also tried to remsh into nodes and force them to do net_adm:ping of its peer (returns 
pong - connectivity is restored), as described here (http://www.ejabberd.im/node/3164) but 
this did not help either (btw, is this expected behavior?)

I can fix the problem only by restarting a rabbit node on at least one host.

Is there a way to restore cluster connectivity by remsh'ing into a node and calling some 
command? I don't want to restart nodes, or even restart app only if at all possible. Has 
anybody experienced problems like this?



Thanks,
Dmitriy




More information about the rabbitmq-discuss mailing list