VSAN Part 17 – Removing a Disk Group from a Host

[Updated: 17th Feb 2016] Following on from my recent post on how to reclaim disks that were previously used by VSAN, I was asked how one can remove a disk group from a host that is participating in a VSAN [5.5] cluster. This is quite straight forward, but there is one minor caveat and it relates to whether the VSAN cluster has been setup in Automatic Mode or Manual Mode. If you want to learn more about the behaviour of the different modes, you can read up on it here.

To begin with, I suggest [strongly recommend] that you place the ESXi host with the disk group that you wish to remove in maintenance mode. While this is not a necessary step to deleting a disk group, I would think that most administrators might like to move the virtual machine components currently in this disk group to other disk groups in the cluster before deleting the disk group [and VMware strongly recommends that you do]. If you don’t do this step, and evacuate the data, you may be left with degraded components which are no longer highly available while VSAN re-configures these components. [In a worst case scenario, if you do not evacuate all data out of the diskgroup, there is the possibility of losing data for reasons that I will outline shortly]. Therefore I recommend entering maintenance mode first, and evacuating the data in the disk group before deleting it. Entering Maintenance Mode is done by selecting the correct ESXi host and then clicking on the maintenance mode icon in the Disk Management section on Virtual SAN in the vSphere web client (third icon from the left):

Enter maintenance modeYou are now prompted with various option related to maintenance mode. You can read more about these options in another post I did on maintenance mode here.

mmode optionsMy recommendation is to migrate all of the data from this host. However, as I said, this is not a necessary step, so you can decide to skip it if you wish. [This sentence was recently brought to my attention. Since this post is 2 years old, we’ve learnt a lot about behavior and best practices during this time. Although the screenshot above is showing Ensure accessibility, I would now reinforce my recommendation and if at all possible, do a full data migration of all data from the host that is being placed into maintenance mode. We have had a few examples (enough to make us cautious) of customers using “ensure accessibility” without realizing (a) that there was an existing issue/failure in the cluster or (b) that there was already a host in maintenance mode or (c) something else going wrong during the maintenance period. You have now inadvertently introduced multiple issues/faults, impacting VM and data availability. Therefore always do a full data evacuation when decommissioning a VSAN 5.5 host. Yes, in VSAN 5.5 its a pain that we cannot do individual disk group and disk evacuation like we can do in later releases, but this should be another good reason to upgrade.]

Once this step is completed, [and all the data is evacuated from the host], the host is now in maintenance mode, and you should be able to remove the disk group. However, the icon to remove the disk group is not visible in the Disk Groups view.

3. no remove the disk group iconThe reason for this is because the cluster is still setup in ‘Automatic’ mode, which means that if we removed a disk group, VSAN would immediately claim the disks once again. So you will need to place the cluster in manual mode for the “Remove the disk group” icon to appear and thus remove the disk group.

4. manual mode

When the VSAN is placed in Manual mode, return to the Disk Management view and you should see that the “Remove the disk group” icon (it has the red X) is now visible when you select the disk group on the host which is in maintenance mode:

5. Remove Disk groupYou can now proceed with removing the disk group.[Note that this procedure is much easier on VSAN 6.x where new functionality exists to decommission individual disks and disk groups via the UI, and does not require a full host to be placed into maintenance mode and complete evacuated of data/components.]


Comments are closed.