Unable to created Desktop Pool from NEW snapshots

[pablocabron]

Hi, I really hpe someone can shed some light here....

I have 4 master VMs, and they're all displaying the same behaviour. Up till now everything has been fine, with no issues. We're working on Windows 10 1809. Horizon 7. VSphere.

What's happening is this... I can create desktop pools with any snapshot I took previously. i.e. up to a few days ago. So there's no problem with account permisions, rights, etc. That all still works. What I can't do, is fire up one of those VMs, make a change, shut it down and snapshot it, and then create a NEW pool from that snapshot. What I get is 'UNKNOWN ERROR: FATAL Customization appears to not have suceeded'. Which is strange as we aren't using any customization?

One thing that has changed is that when I turn the master VMs on they report that the trust relationship has failed. Now I've looked into this and I've been in and changed the reg key to stop the password change. I've also done the dance of removing/re-adding the machine to the domain. I've also tried resetting the computer account in AD. I even tried removing the computer account, before re-adding it to the domain. Nothing seems to satisfy the Horizon. It flat-out refuses to create any new pools from a new snapshot, but will quite happily build pools from the existing one (i.e. the very same snapshots that report they've fallen off the domain when I turn them on!).

Can anyone help with this? Please let me know if there's more information I can give.

Thanks!

[RoderikdeBlock]

What kind of changes did you make in the VM before created the snapshot? Did you update VMware Tools or Horizon Agent?

[sjesse]

run ipconfig /release and Ipconfig /flushdns before creating the snapshot. If that doesn't work you may need to restart horizon

VMware Knowledge Base

[pablocabron]

I haven't tried ipconfig /release as there's no DHCP. I have flushed the DNS on one of the attempts, yes. But that didn't seem to help. We have also tried restarting the Horizon servers.

[pablocabron]

Very few changes were made. I've even tried making almost no changes apart from the process of re-adding the machine to the domain, and still no joy. The Horizon agent hasn't been updated at all since installation. The VMWare tools were updated a couple of weeks ago, but it's been working since then up until now.

[sjesse]

I haven't tried ipconfig /release as there's no DHCP. I

I haven't tried ipconfig /release as there's no DHCP.  Also for the restart you need to make sure you restart vcenter while the connection servers are down speciically. One issue that occurs is the connection server cache the vcenter session information, which restarting vcenter while the connection servers are down forces this to be updated.

[pablocabron]

Ahhh, OK. That's interesting. I'll give that a go next time. I'm not back in there until Monday now, but thanks for that. I appreciate any and all information that anyone has!

[sjesse]

Why don't you have dhcp, how are the desktops getting ips? Sorry looks like my last response had a copy and paste fail? I've seen this issue where the ip address was saved in the snapshot and it was screwing with the cloned desktops getting new ips. I'm at a loss at thinking of options other than dhcp right now, as each needs its own ip.

[pablocabron]

Sorry, you're right. There is DHCP for the clones. I was thinking about the Thin Clients that we use to access the virtual desktops. They're all static. My mistake.

[hschimpf]

So we've had multiple issues in the past with master images that were joined with the domain. Going back and forth in snapshots can skew the trust relationship between the machine and the domain. The customization that is done during deployment is essentially the VM being prepared for use i.e. changing the VM name to whatever you configured the pool for, joining the domain etc.

When you boot up the master image for changes, you should always run OSOT to revert possible changes to the registry that are not desired to go into the new image and prepare the master for deployment. There is a powerful script framework that helps with image generalization before deployment. You should look up Base Image Script Framework (BIS-F) for this. Just google BIS-F and the first result should be it. This tool lets you run all manner of scripts when shutting down the master. We've configured ours to delete the admin profile and clean up event logs for better overview. It can also use OSOT templates to make sure that unintentional changes in the registry get ported into the snapshot. Since we started using this we were able to get a very consistent master image.

Best regards,

Raetke

[pablocabron]

Thanks for your reply! I will look into this further. We did run the Optimization Tool during the creation of the master image, but I haven't thought about re-running it since or during creation of further snapshots.

[hschimpf]

When updating/changing the master image we always use BIS-F when shutting down before snapshotting. As I said it cleans all sorts of caches, logs and unnecessary data. I can only encourage everyone to do this as you can never expect what Windows does in the background. I'm currently troubleshooting 1909 for a new master image and apparently so much has changed in the user creation process that our login times have gone from ~50 seconds to ~120-160 seconds and I can't figure out why.

Best regards,

Raetke

[pablocabron]

HI, new information. I've found an event log that states the following:

The following information was included with the event:

BROKER_PROVISIONING_PUBLISH_IMAGE_FAILED

Image Publish initiated by InitialPublish for Pool pool-it-dev failed for Image [VM=/VxRail-Datacenter/vm/A-W10-1809-01-D, Snapshot=/VM Snapshot 12%25252f10%25252f2019, 2:17:22 PM/VM Snapshot 12%25252f12%25252f2019, 3:21:13 PM/VM Snapshot 2%25252f3%25252f2020, 12:04:30 PM/VM Snapshot 2%25252f12%25252f2020, 2:52:33 PM/VM Snapshot 2%25252f17%25252f2020, 10:11:44 AM, State=PUBLISHING]

Attributes:

PoolId=

Severity=ERROR

Time=Mon Feb 17 12:04:42 GMT 2020

NgvcOp=InitialPublish

Module=Broker

Image=Image [VM=/VxRail-Datacenter/vm/A-W10-1809-01-D, Snapshot=/VM Snapshot 12%25252f10%25252f2019, 2:17:22 PM/VM Snapshot 12%25252f12%25252f2019, 3:21:13 PM/VM Snapshot 2%25252f3%25252f2020, 12:04:30 PM/VM Snapshot 2%25252f12%25252f2020, 2:52:33 PM/VM Snapshot 2%25252f17%25252f2020, 10:11:44 AM, State=PUBLISHING]

Source=com.vmware.vdi.desktopcontroller.NgvcHelper

Acknowledged=true

The specified resource type cannot be found in the image file

The description for Event ID 105 from source VMware View cannot be found. Either the component that raises this event is not installed on your local computer or the installation is corrupted. You can install or repair the component on the local computer.

If the event originated on another computer, the display information had to be saved with the event.

Some googling found

https://kb.vmware.com/s/article/2144938

From reading seems like an installer has not finished correctly in the image. I seem to remember there is a command to run to complete any pending installs? .... Investigating.

[milindng]

I am also facing the same problem what you mentioned in your question. I am spending my daily time to troubleshoot this.

Although I have tried with new provisioning of master image of Windows 10-1809 and successfully created desktop pool as well as updated desktop pools. But when I updated this master image for windows patching then with that new snapshot on same master not allowing to create new desktop pool as well as not allowing to update existing one. I am using Horizon 7.7. When I push this new snapshot to any pool for updation I get the error message like "Published Failed".

I have prepared 2 to 3 separate master images to cross check but with every master image I can create pool with 1st snapshot without windows update. In case I install any another software and push image with new snapshot, it shows error  "Published Failed"

I have some master images from last year too where I can create pool from old snapshot but if I try to create another pool with new snapshot, after pushing it to pool, it will give error "Published Failed"

[milindng]

Follow steps:-

1.  Unjoin and rejoin existing master image and push it again. Prefer to rename master image name.

2. If this is not working, then you can make a anther
    Persistent VM from this Master Image. Rename the VM name and unjoin, rejoin this VM to domain.

3. Try to push this new master image to existing desktop pool to get it updated. For me this solution has been worked.

4. If this is not work then make a fresh Master Image , update windows and try to push it. This is little time consuming but
    I think this can also work as for me this solution worked.

i am facing same problem from many days and I used to troubleshoot in this ways as after each Windows  update same problem coming . 🙂

[pari2k3]

Give a try by setting the below on master VM

1. Ensure that "C:\Windows" is included in the PATH environment variable
2. Update the BootExecute with the absolute path -C:\Windows\System32\vmware-svi-nga.exe