VCSA on a VVOL? Is this supported?

[Raudi]

Hi,

i just started again with playing around with VVOLs, i installed my VASA appliance on a local small datastore and all is working fine.

Except one thing:

I can't vMotion my VCSA from one host to a other.

In the Web Client the vMotion freezes between 40 and 50%, because the whole vCenter isn't pingable.

If i look on the destination esxi host, in the Web Client the VCSA is on the host but shows switched off, on CLI in the task list the VCSA is running on the destination host. On the source host is nothing. I waited a few minutes but nothing happens. So i killed the vmx task of the VCSA. After that the VCSA disappered from the hosts inventory. Then i used the datastore browser to insert the VM to the inventory and was able to startup the VCSA normally.

So, before i go deepter with my research and creating a case for this problem. Is it supported to run the VCSA on VVOL's?

Kind regards

Stefan

[NicolasAlauzet]

Hi Stefan,

I was doing some research and I couldn't find a statement that says it is not supported. Even as vCenter manages all the vvol construction it looks like it is supported or should work anyway (its another VM at the end)

Did you have the issue with only one host or did you try again the same to other host and encounter the same issue?

(There was and old issue related to sizes of disks that generated errors in vmotions in the past. It is worth a try: Check the disk sizes for your vCenter, and make it to a round number, like 50GB, no 49,39GB for example)

Do you have other things running in vvols? Did you test other vms migrations?

[Raudi]

Hi Nicolas,

the issue is in both directions on my two hosts. And i had no issues in the past, where the VCSA was on a NFS 4.1 datastore.

I will start with doing some research in the logs...

Other vm's on the VVOL datastore can be migrated without problems.

[NicolasAlauzet]

Ok!

Check the disk size, I know sounds silly but was a known issue and many people were having issues migrating vms to vvols that doing that workaround was fixed.

[Raudi]

The second disk is 4.67 GB, i will take a look and report...

[Raudi]

I made some research and the problem with the disk file size will result in a problem during a "storage" vMotion, when migrating from VMFS to a VVOL datastore.

In my case i can't do a normal vMotion between two hosts... This is different.

In the logs i can't find errors, here i looks that the vMotion was finished.

[Raudi]

I have tested a little bit more.

When i migrate other vm's which are on a VVOL datastore, then the migration stuck a 87% for perhaps 30 sec. and 8 to 10 pings are getting lost when i'm pinging the VM.

When i migrate a VM on a NFS 4.1 datastore, on the same storage, i have no or one ping lost, so here is this normal...

I think i must open a case...

[NicolasAlauzet]

Yah, seems to be the best option. =(

[nachogonzalez]

Hey, hope you are doing fine:

Although it's supported to have VCSA on a VVOL, I don't think it's ideal. (same goes for VASA)
If for some reason you loose the VVOL you will loose both VCSA and VASA provider.

Just my 5 cents

warm regards

[Raudi]

Yes, i got already a info from NetApp that the VCSA on a VVOL Datastore is no problem. But the VASA can't be on the VVOL, chicken egg problem. 🙂

NetApp didn't found a issue in the logs, they told me to open a case at VMware, this i have just opened...