vSphere deployment – Cleaning up vSAN after failed vCenter deployment

This is a problem I came across while trying deploy vCenter 9 onto a single vSAN node.

Turns out that there is a known issue:

Source: vSAN

So once you have deployed the vSAN you need to remove the datastore again. You can’t delete the datastore from inside the ESX GUI. The option is greyed out.

The answer is to use a couple of esxcli commands:

1: enable the SSH service in the ESX host and login to the host using root.

2: Run the following command:”esxcli vsan storage list”. It  will give you a list of the vSAN devices used for vSAN.

Now you need to remove those devices from vSAN. To do this you grab the VSAN Disk Group UUID from each device.

Then you run this command: “esxcli vsan storage remove -u <vSAN Disk Group UUID>” Replace the <> with the UUID. Like this:

esxcli vsan storage remove -u 523f5c95-1d85-15df-65cf-2e82f67b366f

Once completed for all devices, re-run the first command: “esxcli vsan storage list”, which should now be empty.

3: Next we need to also take the node out of the vSAN cluster.  Run the following command to get the info required to do this: “esxcli vsan cluster list”

You will get a list like this:

Grab the sub-cluster UUID and run the following command:

“esxcli vsan cluster list”

Now use the following command to eject the host from the vSAN cluster:

“esxcli vsan cluster leave -u <Sub-Cluster UUID>

in my case like this:

esxcli vsan cluster leave -u 523d96cf-1b29-bf74-ba1a-0d3098673525

This may take 30 seconds or so to complete and afterwards if you refresh the ESX UI you should now no longer see a vSAN datastore.

I suggest you restart the host(s) afterwards to avoid any “garbage” being left behind.

Following the restart you can retry the vCenter deployment.