microsoft / MSLab

Azure Stack HCI, Windows 10 and Windows Server rapid lab deployment scripts

MIT License

1.19k stars 284 forks source link

mslab 24.04 - Importing hydrated DC VM fails to boot if the VM OS is WS 2025 #589

Closed Karl-WE closed 1 day ago

Karl-WE commented 5 months ago

Scenario: Host OS (Hyper-V): b26085 Windows Server 2022 Guest OS (DC): b26085 Windows Server 2025 mslab version: 24.04.01 - first noticed this in a 23.x release, appears to be related to WS 2025 changes.

Issue: Using WS 2022 as host OS and WS 2025 for DC (2_CreateParentdisks.ps1). DC template can be created successfully. When the hydrated DC is imported for a lab using 3_Deploy.ps1, the imported DC VM will fail to boot.

Repro

Install WS 2022 as host OS
enable Hyper-V
deploy mslab 24.04
Use WS 2025 ISO during 2_createparentdisks.ps1

Workaround: Use WS 2022 as a guest OS

Karl-WE commented 4 months ago

This issue is under Investigation by Microsoft.

Karl-WE commented 6 days ago

related #596 #588

michbern-ms commented 5 days ago

There were two OS fixes that were made to address this issue, and both are present in build 26296.

I ran the full scenario this morning with 26296, downloaded from the Preview center: https://www.microsoft.com/en-us/software-download/windowsinsiderpreviewserver?msockid=1818fad5388d6a5b3e2ceef839376bbb

I used a Server 2022 VM as a host, brought up to current patch level.

And it worked well. I did not need to use the 'NoDehydrateDC' feature - it all just worked.

jaromirk commented 2 days ago

26296 worked for me

Karl-WE commented 2 days ago

Thank you for testing Jaromir

Was this the hydrated DC VM ? What's the OS version of the Hyper-V Host?

jaromirk commented 2 days ago

Yeah, in the screenshot it's DC. Host OS is ws2025. But I think imported dc did not work there, or did it?

Karl-WE commented 2 days ago

I see, the creation of the DC b26296 did not work (2_CreateParentDisks.ps1). Host OS is a b26296, too. Will evaluate this situation with WS2022 / 2025 b26100 as host OS. Remote related also for the AzHCI 23H2 2408 and 2311 (3_Deploy). Maybe something changed in b26296 that these things arise.

Good to hear with ws2025 b26100 as host OS this should be fixed, now you and Michael reporting the same.

Karl-WE commented 2 days ago

Can repro that the host OS plays a role. Testing with WS 2022 as host OS there is no error with the DC. Michael has more details.

michbern-ms commented 1 day ago

Since the repro steps for this issue explicitly mention WS2022, can we close this one now?

588 will continue to track the WS2025 host portion.