VDI on VSAN Upgrade – Part 4: VSAN On-Disk format

In this post I’m going to be performing an upgrade of the existing VSAN On-Disk file system format. This new version is based on VirstoFS technology, which is a log based file system that performs highly scalable snapshot and clone management support. There is also a new snapshot format introducted in 6.0 (v2) called vsanSparse. This is going to provide additional performance in snapshots and clones, which is ideal for a VDI environment. I am most looking forward to this part of the upgrade!

Upgrade VSAN On-Disk format

1. Login to your vCenter via RDP (for me this is Windows, you can SSH if you run the Appliance)

2. Run the CMD prompt as Administratior and run the following:

3. Login with the administrator@vsphere.local credentials when provided.

4. Navigate to your cluster object by using the command line:

5. It is easier if you mark the cluster object variable for ease of use within RVC.


6. Perform some health checks on VSAN before starting, run the commands:



Note: These commands show useful information about each host in the cluster and other useful information. Be sure to read everything and understand the output. Check for anything that doesn’t seem correct. You can also see under the “Status Health” that the disks are on v1 of the file system format.

7. Next, run the following command and you should see a clean output (hopefully):


Note: If you do see any other output, e.g- Inaccessible objects, the issue must be addressed before proceeding with the upgrade. I made a blog post about an issue I encountered. Failing that, do the usual googling and contact GSS if required.

8. Once all checks are complete, run one of the following commands (I ran the first as I’m only 3 node):


9. This process can take a long time as it goes through each host and unclaims/reclaims disks, moves VSAN objects between hosts and formats the disks.

10. To monitor the progress of the upgrade, you can open a separate RVC session and run(and re-run) the following:


11. At the end of the whole process, some objects may not upgrade so you might see the following:

I believe this is because as the last host is moving off data to format the disks but doesn’t quite move everything off successfully before doing so.

12. To resolve this, re-run the upgrade command as per step 8, this will check the state of all hosts and finish off the upgrade of the final objects that got missed.


13. Run a check on the cluster again to confirm:


This was a fun experience for me, especially I had issues before I was able to upgrade – which I blogged about seperatley. During this time I would say I definitely increased my knowledge of VSAN. I know this upgrade process is actually available in the Web Client since 6.0 U1, but I thought I’d do things “the old fashioned way” because I believe it gives a greater insight into the inner workings of what is happening.

The next post in the series will focus on final host and cluster preparation before proceeding with the upgrade of VMware View.

Leave a Reply

Your email address will not be published. Required fields are marked *