[rabbitmq-discuss] Detecting mnesia network partitions in 2.8.6

Travis hcoyote at ghostar.org
Mon Jan 28 22:02:11 GMT 2013


Hello,

We recently discovered that our 2.8.6 cluster is occasionally partitioning
itself.  We're trying to come up with a method to detect this so we can
respond in a more timely manner.

http://www.rabbitmq.com/partitions.html says that doing a rabbitmqctl
cluster_status should show something like:

# rabbitmqctl cluster_status
Cluster status of node rabbit at smacmullen ...
[{nodes,[{disc,[hare at smacmullen,rabbit at smacmullen]}]},
 {running_nodes,[rabbit at smacmullen,hare at smacmullen]},
 {partitions,[]}]
...done.


to give you an idea of any partitions that might be occurring.

2.8.6 has a different output, though:

Cluster status of node 'rabbit at aus-rmq4' ...
[{nodes,[{disc,['rabbit at aus-rmq3','rabbit at aus-rmq4']}]},
 {running_nodes,['rabbit at aus-rmq3','rabbit at aus-rmq4']}]
...done.


So, my questions are:

1) is this partitions output a feature of a newer version of RabbitMQ?
 Which version?
2) Is there a reliable way of detecting a partitioned mnesia database in
2.8.6 that doesn't rely upon tailing the log file and looking for something
like:

  Mnesia('rabbit at aus-rmq4'): ** ERROR ** mnesia_event got
{inconsistent_database, starting_partitioned_network, 'rabbit at aus-rmq3'}

We will be upgrading soon, but we're not quite through with our testing yet
so we can't just flip the version right now.

Thanks,
Travis

-- 
Travis Campbell
travis at ghostar.org
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.rabbitmq.com/pipermail/rabbitmq-discuss/attachments/20130128/5488b764/attachment.htm>


More information about the rabbitmq-discuss mailing list