Inaccessible VM entries in vCenter after an ESXi host crash

Products

• VMware vSAN

Issue/Introduction

When an ESXi host is not shut down cleanly (e.g. hardware crash, power failure...) running VMs on that host will crash, and entries for these VMs may remain on the ESXi host. These entries will be listed in vCenter as inaccessible.

For VMs running on vSAN, there will be a vSAN datastore path listed in the vCenter view.

Environment

VMware vSAN (all versions).

Cause

When VMs are recovered, either by HA or manually, new entries will be registered for them on the ESXi host they are now running on.

In the case of an unclean power-down the old entries may remain on the original ESXi host, and will show in the vCenter inventory view in italicized format with a word like inaccessible in brackets after the path.

 

/vmfs/volumes/vsan:########=########/########-####-####- (inaccessible)

 

Resolution

First, ensure that all VMs in this cluster/datastore are healthy, and that the vSAN object health check in the Skyline Health view is green.

If there is a problem with vSAN object health after an ESXi host crash, open a support case with Broadcom to investigate this.

 

If vSAN data is healthy, and VMs are running, log on to an SSH session on the ESXi host that these inaccessible items are on. (The ESXi host can be seen when the entry is clicked on.)

Run this command:

vim-cmd vmsvc/getallvms

 

At the start of the output, look for messages like "Skipping invalid VM: 1"

Note these numbers.

 

Run this command for each invalid VM entry to remove the reference on the ESXi host:

vim-cmd vmsvc/unregister "number of invalid VM"

 

e.g.

vim-cmd vmsvc/unregister 1

 

When done, vCenter will now show the same entries as "orphaned" instead of "inaccessible".

/vmfs/volumes/vsan:########=########/########-####-####- (orphaned)

 

Right-click on each orphaned entry, and select "Remove from inventory" to clean these up.