Comment 11 for bug 1873265

Revision history for this message
Ken Giusti (kgiusti) wrote :

I'll be first to admit that my knowledge of rabbitmq clustering is not very deep. However I don't believe that there is a way to have oslo.messaging even detect that rabbitmq is in split brain much less recover from it. In fact I'd venture to guess that even the rabbitmq node itself isn't aware it's in split brain, else it would initiate some sort of cluster recovery I would imagine (again, not much experience in this area).

oslo.messaging's fault recovery is limited to connection loss and failover (via configuration of the transport url). It really has no further insight into rabbitmq's inner state - including the health of the cluster.