Changing VMs vSAN storage policy to RAID 6 does not increase free space on vSAN datastore

[syt75013]

Hello,

I am currently changing the VMs vSAN storage policy of our VMs to RAID 6 (they are currently using a RAID 1 policy) in order to gain more available space in the vSAN datastore.

Using RAID 5 or RAID 6 Erasure Coding (vmware.com)

At the moment, I have already changed the policy of several VMs and the vSAN datastore available space does not increase but instead is decreasing.

As an example, I have changed the policy for 10 VMs in RAID 6 on December 7 at around 11:30 AM, after the objects resync we can see that the Available Capacity did not increased.

RAID 6 should use less storage capacity compared to RAID 1 according to VMware's docs but since I've changed the policy to RAID 6 for several VMs, my vSAN datastore Used Space went from 65% to 71% (so the Available Space decreased actually).

So I'm wondering why I am not gaining more available space in the vSAN datastore with RAID 6 policy?

[kastlr]

Hi,

maybe there're still some resyncs in progress.

Changes to a policy temporarily increases the needed storage.

In your scenario vSAN will keep the RAID 1 protection until the RAID 6 is build and fully synchronized.

So for 1 TB usable it will consume up to 3,5 TB RAW (2 * 1 TB for RAID1 + 1.5 * 1 TB for RAID6) until it's fully synched.

Than it will free the 2 TB used by the initial RAID 1.


When you're running out of space, check and enable vSAN Space Reclaimation as it's disabled by default.

You should than take proper action within each VM to free up unused space (i.e. using sdelete on Windows Guests)

Regards,

Ralf

[syt75013]

Hi kastlr,

Thanks for your feedback.

From the Cluster > Monitor > vSAN > Resyncing objects menu, it seems the objects resync has completed (the resync has been completed 2 days ago).

So that should mean the initial space used by the RAID 1 has been freed now right?

But I am not seeing an increase of the Available Capacity of the vSAN datastore. Actually, it has decreased if we compare the situation to before the policy modification.

[TheBobkin]

@syt75013 Can you confirm that none of the VMs you changed policy on are/were running on snapshots at time of the change? Don't go by snapshot manager, check it properly via right-click VM > Edit Settings > Hard Disk > check path is to a base-disk (e.g. not MyVM-000001.vmdk) - mentioning this as if you change storage policy on a VM it only applies it to the last disk in each chain (e.g. just the snapshots if they are present).

You mentioned "10 VMs", more granular specifics would be helpful e.g. they could have 60TB vmdks or they could have 6GB vmdks, there is big difference how much space is expected to free based on this. Can you confirm the VMs/Objects have a thin-provisioned (proportionalCapacity=0) storage policy applied?

If you like you can PM me (as this is a public forum and host/VM names shouldn't be shared) the output of the following run from any node in the cluster and inform me of the VMs in question and I can check these:

# esxcli vsan debug object list --all > /tmp/objout.txt

(assuming hosts are on 6.7 U3 or later, omit the '--all' if on a older version)