Controllerless Networks

Occasional Contributor II

'Split Brain' situation / multiple masters (VC)'s in a cluster after a (partial) power outage



today we had an issue in one of our IAP clusters. This cluster has ~50 AP's on diferent switches. It usually works fine but today we had some power outage issues and some switches (including the IAP's because they are POE-powered) have rebootet.

Finally the problem: I can't say why (because the switches are all cross-interlinked) but we ended up with two separate clusters each having it's own master / VC. This also resulted in an IP conflict (yeah, both VC used the same configured VC address) and in lots of authentication issues because the radius requests never made it back to the right VC.


Is this normal? Shouldn't two separate clusters merge if they see each other or is the 'master / VC' election only made during bootup (or when the current master / VC) get's lost?

We solved this by rebooting one of the cluster-groups and all ap's joined to the other VC again and everithing was fine.


... We even saw to asterix in our airwave for this branch, so also both master's have reported to the airwave ;)


is there anything we could do to prevent this to happen again?


Thanks & Bye,


Search Airheads
Showing results for 
Search instead for 
Did you mean: