Hi everyone,
Since an unwknown amount of time, when users are logging off their instant clone VDI, the VDI is no more deleted.
In the vCenter the action is :
- Revert Snapshot
- Reconfigure Virtual Machine
- Power On Virtual Machine
No more delete operation.. It's good because it let us the history for each VDI but in case of hardware problem/changes, a recompose didn't resolve the issue.
Is this a new "feature" ?
Horizon 8.4 build 19446835 Version 2111
vCenter 7.0.3
Yes, this is the expected behaviour and the VMs are no longer deleted on logoff.
huh? I still see them being deleted.
There is a setting in the pool that lets chose if the delete or not, it was added in 7.9 I beleive. I can't find the vmware kb right now, but control upp has a link that explains it
https://www.controlup.com/resources/blog/entry/vmware-horizon-instant-clones-controlup-a-prime
Hi Everyone,
Thank you for your answer/feedback.
@sjesse Nop, these option are not visible/accessible when creating a Instant Clone pool with Automatic assignment.
@Jubish-Jose This is really Expected? From vmware documentation it's still expected that VDI are deleted after a reboot or user log off opperation. Instant-Clone Desktop Pools (vmware.com)
Did you have any "new" documentation ?
The documentation that you shared is for Horizon 7. Please see the Horizon 8 documentation here: https://docs.vmware.com/en/VMware-Horizon/2303/virtual-desktops/GUID-D7C0150E-18CE-4012-944D-4E9AF5B...
I provide the Horizon 7 documentation because it's the last one that describe the instant pool functionality process.
I already read the Horizon 8 documentation but did not found any related information about a change in the instant clone creation process as well as any documentation about the actual working mechanics instant clone pools.
The only tiny related info is about the Smart provisioning that is automatically enabled in the Horizon 8 version but it's not related to the delete feature...
The delete on logoff was removed in Horizon 8 along with the removal of View Composer. This was to bring parity with linked clone features. The functionality is not mentioned in the Horizon 8 documentation means that it doesn't exist. They could have explicitly mentioned about that change, but they didn't, and I am not sure why.
If you are looking for working mechanics or more technical details about instant clones, please refer to VMware Tech Zone articles (techzone.vmware.com). The general documentation always provide the basic details.
Hi @Jubish-Jose
I think that even the first version of Horizon 8 had the delete operation when restart because I observe this behaviour only now despite I have multiple Horizon infrastructure.
I agree with the fact that this change need to be clearly indicate in the non technically document when presenting the solution.. A technology that change his process without any info
As impact I will put here two example :
1) I have a customer that had sometimes a problem with the vmwaretools/Nvidia Driver on the VM. In the past, after a restart or logout of the user, the VM was completely recreated and the problem was "automatically" resolved. Now the same VM is juste restore from a snapshot without doing anything on the hardware aspect so that when a user logout, another user connect and get the problem on this VM.
2) - Need to be verified - When changing a hardware setting on the master image, we cannot only wait that a user logout like before but we need to delete the VM's. The process is not as smooth as before.
At least it would be nice to have the option to re-activate this function.
@bjohn Are you in Horizon 7 ?
Whether a VM is deleted or only the snapshot is reset depends on the pool settings under Provisioning Settings
Can you show us your settings?
I'm on 8 (2209).
If you select dedicated (vs floating) the Refresh OS Disk After Logoff option is available.
It should work, I'm on 2111.1 at work, I forget what my lab is. If you look at the documentation for 2212 they mention it as well
Worksheet for Creating an Instant-Clone Desktop Pool (vmware.com)
Thank you for the link to the documentation, I was pretty sure that the default behaviour has not changed with Horizon 8.
I will check for the options but I already squeeze the entire Admin console without sucess at the moment.
What is strange is that there are no errors during the refresh or when a user log off.. Like it's a normal behaviour to not delete de VM.
Have you played with the adsi edit tool to modify the ldap database horizon users at all? Their is an audit setting I think you can turn on that prevents vms from being deleted. I can't find the setting right now.
That's exactly what i'm searching right now 🙂
Vmware support asked me somes steps for troubleshooting when I had some AMD/NVIDIA related problems.. but I didn't find settings changes.
Will update as soon as I found the settings modified.
As as said, this is an intended change and not a bug, at least for a VGPU enabled VM. However I think the beaviour will be the same for a normal VM, I can test if required.
Regarding the examples that you mentioned:
1) I have a customer that had sometimes a problem with the vmwaretools/Nvidia Driver on the VM. In the past, after a restart or logout of the user, the VM was completely recreated and the problem was "automatically" resolved. Now the same VM is juste restore from a snapshot without doing anything on the hardware aspect so that when a user logout, another user connect and get the problem on this VM.
A refresh should do the same thing. When a VM is refreshed, it will be as good as the new VM which was provisioned first. If you can share the exact issue that you are facing, we can see how to resolve it.
2) - Need to be verified - When changing a hardware setting on the master image, we cannot only wait that a user logout like before but we need to delete the VM's. The process is not as smooth as before.
Not sure what you mean here. Any change to the master VM needs to be published and pushed to the VMs and you have the choice of waiting for the user to logoff or do a forced disconnect. Also note that in the case of a push image/resync, the VM will be deleted and a new VM will be created from the new snapshot.
You can request for a feature if you think this will be useful for your environment. https://kb.vmware.com/s/article/1002123
Regarding the ADSI edit that @sjesse mentioned, the policy was pae-VmDeletePolicy=deleteonuse if I remember correctly and it was removed as part of this change.
please provide documenation as I beleive your wrong, I have a 2111 installation and I see in vcenter the vms are actually being deleted. You shared the 2303 documenations
In
it talks about deleting the vm, with a last updated date oon the page being 03/17/2023. If you look at
this is what I was talking about, the forensic hold. which also states when its enabled the vm cannot be deleted but one you release the hold the vm is deleted
Si I'm not sure what your referring to. I'd suggest the OP open a SR and get an offical answer from Vmware as getting conflicting information here is not going to get the issue resolved.
An to be clear I'm happy to be wrong, I just would like to see it written somwhere, as I've read most of the release notes past 2111 and I haven't see it. I'll probably install 2303 somwhere later and look around.
@Jubish-Jose What about the documentation link that have provided by @sjesse ?
Concerning the setting, I already found this "deleteOnUse" but it's correctly configured so it's seems not to be this one or maybe, as you said, the setting is not functional anymore but again, no information about that.
How big are your desktop pools, there is a smart provisioning feature that may be coming into play. Its probably the the no parent instant clones that don't delete after logoff, because the provisioning time is a bit longer then the parent based instant clonese