Invalid VMs and inaccessible vSAN datastore after power outage or network partition

Products

• VMware vSAN

Issue/Introduction

After a power outage, environmental maintenance, or a network partition event, you may observe the following:

• Virtual Machines (VMs) show as Invalid in the ESXi Host Client or vCenter Server.
• vCenter Server is inaccessible or 'Not Responding' (especially if located on the vSAN datastore).
• The vSAN datastore appears empty or unavailable.
• ESXi hosts may experience hostd service instability or crashes.
• Skyline Health reports a Network Partition.

Environment

 

• VMware vSAN 7.x
• VMware vSAN 8.x (including 8.0 Update 3)

 

Cause

A vSAN cluster requires a functional Unicast agent list for host communication. This list can become empty or corrupted following an ungraceful shutdown (power outage). Additionally, physical network failures (e.g., faulty 25 Gb NICs or cables) can cause network partitions that prevent the cluster from forming, even after power is restored.

Resolution

 

• Log in to each ESXi host via SSH.
• Verify the Unicast agent list status by running: esxcli vsan network list
• If the list is empty, follow the instructions in [KB 326427] to manually recreate the Unicast agent list on all impacted hosts.
• Confirm the cluster has formed and all members are present: esxcli vsan cluster get Check the Sub-Cluster Member Count matches the total number of hosts in your cluster.
• Inspect physical network connectivity and Skyline Health for network partitions. If a specific host or NIC is identified as faulty, isolate the interface to restore stable communication.
• If hostd is unstable, restart management agents or reboot the affected ESXi host.
• Once the datastore is accessible, reload any invalid VMs: vim-cmd vmsvc/reload <Vmid>

 

Additional Information

 

• https://knowledge.broadcom.com/external/article/326427
• If the issue persists, see . Scroll to the bottom of the page and click on your respective region to speak with a representative.