Forum Discussion
Azure Virtual Desktop - Black Screens on logins - What we've tried so far
- We are ramping up our users on Win 11 and so far no issues with regular monthly KB patches on Marketplace 23H2 image. Have max 4 users on per host, will be upping to 8 next week. This is over 20 hosts.
We have the issue still pretty bad on Windows 11. We were told it has to do with our State Repository database filling up. We changed our value for how long the DB should fill up until purged at MS advice. Things looked good for about 3 days and then it all came back. The DB purges on reboots, so I dont know if the DB is just filling up during the day while VMs are running and causing the issue or it its something else.
We arent on 24H2, but we are on 23H2...does 24H2 Make a differenceJust chiming in here with three notes in the hopes that they help other people.
- I'd like to echo what other people have noted: replacing your session hosts with Windows 11 24H2 DOES seem to resolve the issue. As a matter of fact, I have never seen this problem with Windows 11, only on Windows 10. I only started noticing this problem (this time around -- see point 2) from about October 14 forward. I have some self-service session clearing set up for my environment and I noticed a marked increase in the use of that tool (combined with tickets) from that day forward. I replaced one of my pool's session hosts with Windows 11 24H2 and the difference is measurable.
- I mentioned "this time around" above. I first noticed this problem at the end of 2023, mainly in one of my smaller (but higher-used) host pools. I engaged Azure support as I was at my wit's end. I finally did the same thing there, rebuilding the pool with Windows 11 hosts, and again, the problem disappeared. I knew I'd have to do this to all of my pools anyway, but I didn't have the time yet to do it.
- There was a notification sent out yesterday from Azure Service Health providing more information on the issue. As you can probably guess, it contains what everyone already noticed here, and it suggests https://support.microsoft.com/en-us/topic/october-22-2024-kb5045594-os-build-19045-5073-preview-f307a4b0-f62d-4c28-9062-44207aea55c3 (if you can't go to Windows 11) and also suggests that Windows 11 is less impacted by this issue.
Registry Hive: HKEY_LOCAL_MACHINE
Registry Path: SOFTWARE\FSLogix\ProfilesRoamIdentity
Type: DWORD
Default Value: 00: Don't roam credentials and tokens within the container.
InstallAppxPackages
Type: DWORD
Default Value: 0
Just to add, incase anyone else is pulling their hair out trying different things.
The KB5045594 fixed 100% of black screen issues for us. But we still experiencing Outlook/Onedrive issues.
These registry keys did not help, we have the same issue with Outlook/Onedrive stuck on "connecting" since adding these keys and rebooting the servers.
Our only fix is to delete the credentials from credential manager and logging off/on again every time it happens
Looks like we'll be rebuilding our hosts on Windows 11 images over the weekend 😞- Well here is the Service Health Advisory update....
Workaround: This issue can be reduced significantly by installing this latest patch. An alternative workaround is available but may not be feasible for all customers. The suggested alternate workaround for the issue is for customers to move to Windows 11, as the issue is less likely to occur on this operating system. - The KB504559 has fixed 100% of our black screen issues, I removed the KIR and still no more black screens. We do still have the SSO issues when users open Outlook/Onedrive, this presents in different ways from a need password message with no window, to a stuck connecting message at the bottom of outlook.
Removing all credentials from credential manager and logging off the host pool fixes this every time not matter the issue.
I'm trying the 2 reg keys mentioned above now, at this stage I'll literally try anything. I agree its not an FSLogix issue but if disabling roamIdentity within the fslogix profile fixes it, that's good enough for now, I just need to remember to change them back when "IF" a proper fix comes along.
Will report back in 18 hours - I agree, I've experienced the issue via RDP to a machine so no fslogix there.
In the KB they are talking about appx fixes not fslogix though.
https://support.microsoft.com/en-us/topic/october-22-2024-kb5045594-os-build-19045-5073-preview-f307a4b0-f62d-4c28-9062-44207aea55c3 - I did not see that, but if you could mention it again here, that would be great. If it has to do with the fslogix option to 0 then not an option for us as we use appxPackages quite a bit. You seem to mention that you "enable" the KIR. We might try that, but really Windows 11 is working well so far and we were always planning on moving, so now that the Defender issue is reported resolved, I think we will just focus on that.
- I’ve taken several steps, and for the past three days, everything has been stable. I updated FSLOGIX to the latest version, applied the KIR, then disabled it in the GPO. I also uninstalled both the September and October CU patches and installed the Preview patch KB5045594. Previously, I was receiving complaints from about 3-6 users daily about black screens or issues connecting to Outlook, Teams, Sticky Notes, or any application requiring SSO authentication. So far, knock on wood, everything seems good.
- none of this has anything to do with FSLogix.. whether you're using it or not. Black screens and AAD/SSO/Outlook issues were prevalent.
- Apologies, but just confirming for my own sanity.
You tried KB5045594 ? KB's are not helping, we are progressing further with the below registries deployed across our hotspools and it seems fixing the MS apps issues.
Registry Hive: HKEY_LOCAL_MACHINE
Registry Path: SOFTWARE\FSLogix\Profiles
RoamIdentity
Type: DWORD
Default Value: 00: Don't roam credentials and tokens within the container.
InstallAppxPackages
Type: DWORD
Default Value: 0- you don't need to disable the KIR anymore as per Microsoft's documentation. you can just unlink the policy or delete the policy. https://learn.microsoft.com/en-us/windows/release-health/status-windows-10-22h2#3421msgdesc
If you install an update released October 22, 2024 (KB5045594) or later, you do NOT need to use a Known Issue Rollback (KIR) or a special Group Policy to resolve this issue. If you are using an update released before October 22, 2024, and have this issue, you can resolve it by installing and configuring the special Group Policy listed below.
It would also be good to know what's configured in your fslogix Group policy settings and redirections.xml - Yeah all other KB's are installed, have not removed anything. But did you see the Reg key i mentioned in a previous reply? That seems to be the one that fixes everything for us for now. May not be ideal if you use appxPackages, but as soon as i turn that back to enabled we have M365 app issues and black screens back. We were getting issues daily, but like i said previously today we havent had a single call this last week with the fixes i mentioned.
https://support.microsoft.com/en-us/topic/october-22-2024-kb5045594-os-build-19045-5073-preview-f307a4b0-f62d-4c28-9062-44207aea55c3
This is the preview patch they are recommending. I have not yet tested this.
I have also had success with just running 24H2 Windows 11. (Not ideal for every one).
- Can you share if you have all other regular monthly KB patches? We are currently running all KB's and the KIR is currently disabled. We have 20 hosts with 10 users each and we usually get 1-5 users a day needing to be moved to a new host. Can you clarify if you have the Sept/Oct KB's removed also? We are migrating to Windows 11 currently as that was the plan anyways, but this issue and the Defender Exclusions/Network Protection issue that I think was patched yesterday was making us pause that deployment, but now we just see to many pluses for dumping Windows 10 and are very disappointed with the current state of MS support. I will be attending Ignite and I plan to voice this concern to anyone that can listen and will be using this thread for historical examples.
- We also received same communication a little too late.
Just got this in the health alerts:
TYPE: Incident
STATUS: Active
COMMUNICATION:
Impact Statement: Starting on approximately 23 July 2024, you have been identified as an Azure Virtual Desktop customer using multi-session-enabled Windows 10 22h2 devices who may or may not experience the following symptoms:
• AVD-session host logons experience an extended 10-to-30-minute-long hang at a black screen following user logon
• Office applications including Outlook and Teams experience SSO failures that prevents connecting to backend services or synchronizing data.
Workaround: This issue can be reduced significantly by installing this latest patch.
Current Status: We detected this issue via customer reports upon identifying the aforementioned impact. We are aware of this issue and are actively investigating to explore the underlying root cause and potential mitigation workstreams.
This issue affects enterprise users with multi-session environments due to issues with AppX Deployment Service.
The next update will be provided within 11 hours, or as events warrant.
Only 4 months too late!!!
At least they are now acknowledging that the SSO issue is linked, which they were adamant they were not.- At this point, I dont care if we need incantations, magic dust, cruxes and snake oil. If it works, Im happy 😉
