Split Brain in VM machine on Application Level

[doran_lum]

Hi all,

we have two standalone VM machine which have a heartbeat NIC in between so one VM machine will be active while the other is inactive. During the past month, we have seen several errors on the below on the application where both VM machine become inactive. I check on the VM machine performance stats and I don't see any performance issue. Currently as a workaround, we have left one machine running and would need to bring back the other machine online soon.

Error: /sys$hacontroller/ERROR/v930/NegotiateStage, isTempMaster=true/*** Inconsistent state detected: STANDALONE vs. STANDALONE!*** Split Brain Instance Action for this instance

I have read a number of articles and it seems like Fault Tolerance need to be turn on ?

[antonmajor]

If you have two guest os' that are designed to work as a pair with a heartbeat, stay away from FT.

Investigate your networking setup. See if you can have your heartbeat communication sent over a separate subnet, vlan, portgroup etc. If you have all of this setup, check all your network configurations on all your hosts to make sure they all match. There is a possibility that one gets vmotioned to a host that is connected to a vmnic on a port that isn't on the same vlan.. the guest os triggers the failover.... some time later the vm gets vmotioned back(still seeing itself as primary) and you get the error.... shot int he dark, but might be the case.