This message was deleted.
# citrix-vad
s
This message was deleted.
k
Looks like I might have the same issue
I posted a similar issue earlier today
n
Fuckin hell, I guess that makes me a little happier.
Yeah, I just saw your post.
k
It's actually a colleague of mine experiencing the issue
n
When this happens, I have to run VDA cleanup utility, and then still manually delete C:\ProgramData\Citrix\XenDesktopSetup
k
The customer have 600 WIndows comptuers with RemotePC enabled
n
My admin install finished successfully.
and the VDA is now registered
k
in what context are you running the task sequence? Local admin?
n
I have done so many VDA upgrades over the years, and have never seen this behavior.
When it fails, or when it succeeds?
k
Both
n
Fail: Logged in with non-admin account, PowerShell is launched as Administrator and run in that context
Success: Logged in with admin account, PowerShell is launched as Administrator and run in that context
k
How about the SCCM task sequence, is this running in the local user context?
n
I highly doubt it, but also don't know. That's another team entirely.
I assume it's running under SYSTEM like most SCCM packages, as that install method is showing the same issue that I just posted with my bad manual install.
k
Hmm
n
This is what I get for commenting that I never have VDA upgrade issues here.
k
The /noresume cmdline switch should prevent the runonce registry entry though
n
We don't use that switch.
We do use noreboot, though.
Why would you need the noresume switch? If you do use it, do you need to manually kick off the .exe after reboot?
k
Yes, I think that's the idea
n
How the hell do you use SCCM to upgrade the VDA then without a task sequence?
k
I don't know. I don't have very much experience with SCCM
n
I honestly can't recall how that works. I would expect the /noreboot switch to not require any of that, though.
The finishing of the install post-reboot is likely what's killing this
I did try and manually launch the RunOnce entry after my manual attempt (logged in as non-admin, ran PS as admin and launched .exe), but that didn't do anything. The installer just ran multiple copies of itself (new PID every few seconds) and did nothing until I killed the task
k
I have seen the VDA installer spitting out an exit code of 3010, if I remember correctly, even though it installed correctly
n
In our logs, the final entry in the log file was 'pending reboot', and nothing afterwards
Actually looks like this is happening after prereqs, not even the VDA.
Yep, bombs after the VC++ installer.
πŸ‘ 1
k
Yes, it looks like it happens because of a Visual C++ install
bbl
n
Now I wonder if we have to make that package a prerequisite package in SCCM. That would be so dumb.
r
I had to upgrade c++ first then to the vda upgrade
n
Damnit
This just made all VDA upgrades that much more of a PITA
r
But with sccm it seems to always work best to remove the vda clean up upgrade the pre reqs then reinstall the vda. Might also need a reboot in there
n
I don't see any way of doing it without a task sequence then. I have seen the VDA cleanup utility require 3 reboots to complete - like on my own VM just now.
If that utility could run and only require a single reboot, that would work, but I don't see how this functions properly with that many reboots in between.
Unless the /unattended switch resolves that
And honestly, requiring the VDA cleanup utility for upgrades... woof.
r
In sccm there should be sn exit code if you need to reboot. And yes def needs a task sequence. Its not a simple upgrade due to the prereq requirements. You can roll those out separately then it’s usually a much easier upgrade.
n
Yeah, the SCCM guys sent me a screenshot of the TS. Looks like VDA cleanup is being run 4 times, with 4 total reboots, so I don't think that part of the process is the issue. Getting a few VMs back to 1912 so that they can add VC++ back to the TS post-cleanup to see if that yields different results.
πŸ‘ 1
This is the kind of stuff that scares all of us for the auto VDA upgrades in cloud, BTW.
k
Has in-place upgrade of the VDA ever been good? It seems like there has always been issues
n
Honestly, I never really had an issue with 1912 and earlier that I can recall. 2203 for us has been a mess with upgrades.
If adding a separate step for VC++ doesn't work, I imagine we'll be engaging Citrix support, because there's no way we can upgrade the VDA on our thousands of persistent VDAs any other way.
VDA cleanup + reboot x3, 1912 CU5 install + 1 reboot... VDAs register fine
r
When we were doing tons of remotepc installs the key was getting the prereqs installed before trying the vda. The issues always seemed to be when the vda had to do the prereq upgrades. No prereqs and it worked fine.
d
A little late to this party... Sorry if what I say has already been said, I only skimmed through this whole thread. The VDA cleanup utility provides a return code to know if a restart is required or not, so can be scripted to restart as many times as needed, which I do in my deploy script. I use the /noreboot /noresume flags in my VDA deployment so I control the entire process in my script because Citrix doesn't seem to do things consistently, and depending on what pre-reqs are needed there may be more restarts required. My script manages the entire process. I always use a local admin account where the account is granted admin rights on the local machine during script run and uses Autologon to manage the process. Autologon and Admin rights are cleared at script end.
πŸ‘ 1
n
If adding VC++ as a separate step doesn't resolve this for us, I'm going to suggest using /noresume and having them kick that process off via the TS.
e
Normally C++ out of date
n
@Rob Zylowski: There's no hard requirement for VC++ 2015-2019, right? We're going to attempt to use a newer 2015-2022 package.
πŸ‘ 1
r
I think that would work.
n
Just to close the loop on this, that seemed to work. Here's our new process (via Task Sequence in SCCM): 1. Run VDACleanupUtility 2. Reboot 3. Run VDACleanupUtility 4. Reboot 5. Run VDACleanupUtility 6. Reboot 7. Run VDACleanupUtility 8. Reboot 9. Install VC++ 2015-2022 10. Reboot 11. Install VDA 12. Reboot
k
Wouldn't it be great if we didn't need the VDA Cleanup Utility?
πŸ’― 3
n
david rose absolutely
πŸ˜„ 2
Now that we've got the process down for 2203 CU1, I can ignore everything we've seen and immediately do the same for CU2, lol.
k
Sure thing :P
n
I just told the SCCM team, and I'm pretty sure that I could hear them slamming their keyboard. πŸ˜‚
πŸ˜‚ 4
Just started testing CU2, and all of the installations appear to have failed... until I manually RDP into one of them, and it kicks off in the background. Trying to get the install string they're using, because I don't understand why anyone would need to logon in order for the installs to kick off/continue.
d
Sounds fun. Certainly sounds like they didn't run to completion and are not using -noresume in the command line. I know I script the process using -noresume and do my own checks to rerun the process until it completes successfully or fails and notifies.
n
We aren't using /noresume, correct. I think I'm going to have to get them to change this up, because this isn't working. That said, it's 2023 and Citrix still can't build an installer that fucking continues after a reboot automatically, without having somebody login?
πŸ‘ 1
If you're running SCCM for the installs, would you mind posting your task sequence and commands?
d
I'm not using SCCM. I have a PowerShell script which I have posted a "cleaned" version in the past that I can share, but I have re-written it since then to cut it down a bunch (removed a lot of unneeded steps that removed 6.x for upgrades in the past mostly) and I think I have the VDA Cleanup utility functioning better since then, since Citrix added a return code. If you want either, I can probably spend some time doing a quick cleanse of the current script to share.
n
Nah, it's all good. I had our SCCM team just add a step to run the local path post-VDA install, and added some more reboots. Damn thing is going to be take 8 reboots to upgrade a VDA now... insanity.
πŸ’― 1