AVD host crashes when a user logs on

[bodulaja]

I have a weird situation here. My AVD host crashes when one particular user logs on.

Hosts are Windows 10 22H2 Multi-Session with M365

FSLogix version: 2.9.8171.14983

All windows update patches installed to Jan 2023 CU

Observation:

When user logs in via the AVD client or via console logon, if he is the first to log on, host does not crash.

If he logs on AFTER other users are already on the host, he crashes the host. Host bluescreens with RESOURCE_NOT_OWNED BSOD, with ntoskrnl.exe being the faulty driver.

I have recreated his fslogix profile just in case there was something in his old profile, still no improvement.

This is really giving me a headache because he crashes any host he logs on to which might have 7 to 8 users already logged in. This leads to their profile not getting dismounted properly, thereby creating a new local profile for them when they try to log back in.

I have already raised a ticket with Microsoft and no TTR at the moment.

Spinning up a new host has not resolved the problem.

Will appreciate any insight into this issue.

More information available, including minidump files 

Thanks

[#2]

@bodulaja 

Hi, please see my post from last week: Some Session Hosts became unavailable this morning - Microsoft Community Hub

All last week we noticed random occurances that are somewhat similar to yours.

ie. User logs in to a SH with many others on it, all get crashed out (did not see BSOD however). Same profile lock issue upon login again.

Rebooting host the only way to 'fix'

Tried adding new SH but same behavior. Then all went well for the rest of the day.

Not apparent pattern I could see. I had ticket and only suggestion was to rebuild the image for the agent to install again.

Did that over the weekend and so far so good today (Monday 30th) however I don't have confidence in it at the moment at all.

We tried different new profiles, SH, clients. Keen to hear anything you get from Microsoft on this.

[#3]

Hi Paul,

We are seeing exactly the same issues. These have been going on since November. The one thing we haven't tried yet is to rebuild the image. Why did Microsoft recommend you rebuild the image if it doesn't contain any agent as this gets installed post provisioning?

Is the environment still stable at the moment?

Thanks

Harry

[#4]

@HarryPas 

Well didn't make any sense to me either. But I did it to rule out everything and get them to go further with trouble shooting. Was fine all weekend and today until about 4.30pm when it happened again. Pulling my hair out tbh

[#5]

Just found out Fslogix was out of date. Updated now to the latest version. Will try and simulate the crash again to see if the update has helped.

In the meantime,  Microsoft has gotten back to me saying it seems to be a FsLogix issue and a fix might be available mid-February.

[#6]

Hi Paul,

Thanks for the update. We are also pulling our hairs out! PLease update once you have had a chance to test!

Thanks again!

[#7]

to confirm we are running FsLogix 2.9.8228.50276 and have been for quite some months. This issue has really only popped up last 2 weeks. Will advise if I hear any more from Microsoft. I'm feeding back each time with logs and screenshots etc

[#8]

Hi Paul,

Since just upgrading the FSLogix agent, has the system been stable for the last couple of days?

[#9]

Unfortunately, upgrading FSLogix to version 2.9.8361.52326 did not make a difference. The user just crashed a host with 7 other users already logged on.
AVD agent is already on the latest version 1.0.5739.9000/1.0.5739.9800.
Really don't know what to do next now.

[#10]

Microsoft mentioned there was an issue on the Azure side regarding storage accounts in our case. Although I think it is far from definitive.

In any event, yesterday and today we have had zero issues for the first time in past 2 weeks so.. crossing my fingers.

One thing you might want to rule out is device. early on we found we could replicate it by connecting the user from an HP laptop, but not from a MS Surface. Makes zero sense as clients side should not affect this. but maybe test the scenario

[#11]

Hi @PaulStirling 

We have been stable for the last few days as well with not having changed anything to be honest.  Me thinks Microsoft are not giving the full picture!

[#12]

@HarryPas 

Oh 100%. We have been 10 working days without issue. But suddenly last minute today we had it happen again. Users now know what to do to recover (which is a sad lol ). We ended up closing our Ms ticket it all went around in circles. Feel very unloved.

[#13]

Thanks for the update Paul.....hope it continues to get better!