I have a 2 node HA Cluster. One is down but there is still a (stopped) VM on it. How can I move this VM to the active Node ?
The cluster is using a NFS share there is also the data of the VM.
A PVE Cluster works on the quorum/majority principle. Each node has one vote. A two node cluster is a special situation. If one node is down, you only have 50% of the votes and therefore no majority. This means that certain actions cannot be done like starting a VM or changing the config of a VM. The recommendation is to always have an odd number of nodes in a cluster to avoid a "split brain" situation in which it is not certain if the other half is still up or not.
If you cannot add a third node to the cluster, you can use the QDevice mechanism which provides one more vote to the cluster. You would end up with 3 votes in total and would still have 2 votes if one node is down.
You will need to install the corosync-qnetd service on another machine that is external to the PVE cluster. That could even be a raspberry pie board. Then install the corosync-qdevice packets on the PVE nodes and set up the QDevice. There is a chapter in the PVE admin guide explaining the whole procedure.
For VM to move to another node ( ie node which is up) VM must use the shared storage. if VM is already on shared storage
do pvecm expected 1 and then migrate
This site uses cookies to help personalise content, tailor your experience and to keep you logged in if you register.
By continuing to use this site, you are consenting to our use of cookies.