QuestionQ27

VerifiedManage Workloads in a Multicloud Environment

In a 3-node cluster where all nodes are identically configured, one node is running low on physical resources. An administrator attempts to live migrate VMs off that node to another node in the cluster, but the live migration cannot be performed.

What is causing this issue?

Explanation

VMs configured with GPU passthrough (direct/dedicated hardware assignment of a physical GPU to a VM) cannot be live migrated because the VM's state is tied directly to the physical GPU device on that host, and there is no equivalent physical GPU state to transfer to the destination host. This makes GPU passthrough VMs pinned to their current node, preventing any live migration attempt regardless of available resources elsewhere in the cluster.

Community Discussion

No comments yet. Be the first to start the discussion!