MCS not refreshed at reboot

[Edwin Verstege1709153889]

Hi There,

 

We are using Citrix Machine Creation Services to provision multiple XenApp 7.6 machines.

I Thought that the machines should be in a clean state after a reboot.

 

I tried to reboot the machines from Citrix Studio, trough VMware Vcenter and in the VM itself but all changes on the XenApp server are retained after the reboot. Also tried a reboot schedule on the Delivery Group and fully shutdown the XenApp servers but still no luck.

 

Technical information:

Desktop Delivery Controller: 7.6.0.5026

Citrix Machine Creation Services: 7.6.1.5020 (Feature Pack 1)

VDA: 7.6.0.5026

OS: Windows Server 2012 R2 (including all critical updates)

Hypervisor: VMWare Vsphere 5.5.0, 2183111
(We use local storage on our cluster)

 

Hopefully someone can help me with this!

Is there a powershell command available to force a refresh?

[Jarian Gibson]

If you do an image update do the machines update properly?

[Mark Syms]

If the machines are showing as "no" in the pending updates then the system doesn't think that there is anything to do. There are a few reasons why this may be the case the most likely being that the catalog update process didn't actually complete successfully and so there is nothing to associate with the machine. I think you'll need to do another catalog update and see if that works.

 

Mark.

[CarlStalhood]

If you use Studio to reboot the machine, does it reset?

[Edwin Verstege1709153889]

i tried that, unfortunately de VM won't reset

[Mark Syms]

Are the machines actually showing that they have an update pending (use the Pending Update filter in Studio search node).

 

Thanks,

 

Mark.

[Edwin Verstege1709153889]

Hi Mark,

 

All servers display a "No" in the Pending Updates tab 

[Edwin Verstege1709153889]

Nobody any idea?

 

We are using a Server Operating System, could that be a problem?

[Jarian Gibson]

Are you trying a shutdown and restart from studio and not just the VM console?

[Edwin Verstege1709153889]

Yes i have tried to restart and shutdown the XeApp Servers within Citrix Studio

[Edwin Verstege1709153889]

The Machine Catalog is updated to an ajusted VMWare Template. The "Pending Updates" still listed "No" after this update.

Does anyone have an idea how to troubleshoot this?

[Mark Syms]

Updated to a VMware template? To update a catalog you have to select a snapshot of a VM and allow XenDesktop to perform the update process. Please detail the steps you've actually performed.

 

Thanks,

 

Mark.

[Edwin Verstege1709153889]

Hi Mark,

 

Sorry, i meant VM Snapshot

 

Hereby the steps in detail:

Starting the master VM

Created a tekst file in C:\temp

Shut down the VM and Created a new snapshot

Updated the Machine Catalog with the new snapshot

Choose "on next shutdown" as rollout strategy

Awaited till the full process was finisched and checked the "Pending Update" status

[Dave Wormall]

Hi Edwin this sounds exactly like the problem I posted up yesterday that we are having http://discussions.citrix.com/topic/368032-mcs-not-updating-machine-catalogue-servers/ 

 

Id be interested to find out whats causing this. 

[Jos van der Haar]

Im experiencing the same problem on XenApp 7.6 Server 2012 R2.

I have 3 Machine Catalogs. 1 that works (Resets on vm poweroff/on) and 2 that does not work.

All 3 says "Discard" in Persistent User Changes. 2 of the 3 catalogs is from the same master image snapshot.

 

I've noticed that in vSphere the working Catalog have checked the "Non-Persistent" flag, the 2 nonworking Catalogs has not. Maybe this is the reason?

 

Where can i debug futher?

 

Thanks.

[Mark Syms]

That is exactly the reason and I would suspect external interference has caused things to not be configured correctly.

 

When XenDesktop creates non-persistent machines on VMware it uses independent non-persistent disks on the VMs. If you have catalogs that don't have this set then they will not work as clean on boot machines.

 

Mark.

[Edwin Verstege1709153889]

Hi Jos,

 

The created VM's by the machine catalog creation wizard have two disks in VMWare.

The first disk is a persistent disk and the second disk is a non-persistent disk.

 

Did changing the persistent disk to non-persistent solve the problem?

[Jos van der Haar]

Just to clarify:

The working Machine Catalog has the following disk mode settings:

Disk 1 - Identity Disk - Independent Persistent

Disk 2 - Deltadisk (Linked clone from basedisk/masterimage) - Independent Nonpersistent

 

The NOT working Machine Catalog has the following disk mode settings:

Disk 1 - Identity Disk - Independent Persistent

Disk 2 - Deltadisk (Linked clone from basedisk/masterimage) - Independent mode checkmark is unchecked = Neither persistent nor nonpersistent 

 

I did not try to change the Independent mode. At least i cannot do it while the vm is running.

The MCS should already have done this though.

I'll try to create a new machine in the catalog and change the setting to nonpersistent while its shut down, and write back.

[Mark Syms]

MCS will have done this.

 

Either,

 

 * You are actually looking at a persistent catalog which would be built exactly as you have described, or

 * Somebody has fiddled with the VM setting externally to XenDesktop.

 

Mark.

[Jos van der Haar]

Ahh. new information for me. I was not aware that you can actually choose to create a persistent catalog.

How do you do that?

 

I dont think someone has fiddled with settings, because, when i create a new catalog, it's also created as "persistent".

 

I've not had the time yet to test futher.

[Mark Syms]

Jos,

 

Can you give me a bit more information?

 

What type of machine are you creating, desktop or server?

What hotfixes and version do you have? Specifically do you have 7.6 HF2?

 

Thanks,

 

Mark.

[Jos van der Haar]

I have now tried to manually set the independent mode to nonpersistent.

This does not fix the problem.

When i poweroff/on they do revert, but only to when i made the change. 

After i made the setting to nonpersistent, i tried to update the catalog.

I had 1 server powered off, and 1 powered on. 

I've told in the update process, that i should reboot the servers immediately.

The powered on server sure do rebooted, but it did not use the new basedisk. Neither did the powered off machine.

The workaround i have now is to delete the machines from the catalog, AD and Vmware, and to create new ones.

When i create new machines in the already updated catalog, the new machine uses the new basedisk.

[Jos van der Haar]

Jos,

 

Can you give me a bit more information?

 

What type of machine are you creating, desktop or server?

What hotfixes and version do you have? Specifically do you have 7.6 HF2?

 

Thanks,

 

Mark.

I'm creating Servers (Server 2012 R2)

The controllers are vanilla.

I applied 7.6 HF2 today though, to see if it made any changes. It did not.

I have yet to try and see if it works creating a new catalog.

 

Any tips?

[Jos van der Haar]

I have yet to try and see if it works creating a new catalog.

I just tried creating a new catalog.

The update 7.6 HF2 did not resolve the issue.

It still creates persistent disk.

 

How can i debug the creation of the catalog? 

[Mark Syms]

Sorry, I should have been clearer in the earlier email. Specifically HF2 has a bug which causes this behaviour and will need a new update.

 

Can you get the output from the following PowerShell command

 

Get-ProvScheme

 

Specifically the CleanOnBoot property.

 

The HF2 (and possibly earlier fix versions but not the 7.6 release version) of Studio have a bug and set the catalog to non-persistent and the ProvisioningScheme to persistent.

 

 Mark

[Jos van der Haar]

I think you're on to something! :)

On the working Catalogs CleanOnBoot is True

On the non working Catalogs its false.

 

How do i fix it? I guess using Set-ProvScheme? Can you help me with the syntax?

And do i need to update or delete-recreated the VMs for it to take effect?

 

Thanks.