- Posted In
- Red Hat Virtualization
RHEV3.0- High Availability during an manual reboot/shutdown/ hard reboot of Hypervisor
I have an RHEV Hypervisor 6.2 configured and using RHEV 3.o manager. I have high availability enabled to all the VM’s. now when I reboot one node, I see VM’s going in to unknown state. I don’t have power management configured. So does high availability don’t work on reboot/shutdown/hard reboot of Hypervisor ????
> So does high availability don’t work on reboot/shutdown/hard reboot of Hypervisor ????
Usually manual reboot and shutdown should be done after putting the hypervisor to maintenance mode which will trigger migration of vms to other hypervisors.
High Availability will not work without power management. Without fencing the hypervisor there is no way to make sure that the vm will not end up running on both hypervisors which would cause filesystem corruption and make your vm corrupt beyond recovery.
Since RHEV uses the Red Hat Cluster Suite, power management is key to all the HA and clustering features. Without it, you have to perform the fencing commands manually.
This should be mentioned in the admin guide of course, but in short, if you have no power management (or for some reason it is not functional), you always have the option to rightclick the host that went down and select "Confirm Host Has Been Rebooted". This will allow RHEV-M to proceed as if the host has been fenced.
You have to be careful with this, since at this point you are manually overriding the fencing flow, and if the host is actually still up, you can lead to a splitbrain this way (this is also mentioned in a warning in the GUI when you click this option). So be extra careful with this, and when clicking "confirm host has been rebooted", first, take the time to actually confirm the host is down or rebooted, and is not running any VMs. This is very important.