BootWiz.exe crash when compiling WinPE preboot image

[deemacgee]

Hi all,

Bootwiz has just recently begun to crash on every attempt to recompile WinPE preboot images. Unfortunately, this has come at an absolutely awful time because we had intended to use PXE-based WinPE to repair the damage done by the recent Crowdstrike outage.

The last few lines in the bootwiz log file read as follows.

2024/07/22 07:16:06 I: .......<CPlatformWinPe2::WimScratchSpace - returning 0 (BDC_SUCCESS)
2024/07/22 07:16:06 D: ...... Product is detected as NS
2024/07/22 07:16:06 I: .......>CPlatformWinPe3::UpdateWimFileRegistry
2024/07/22 07:16:06 I: ........>CPlatformWinPe3:UpdateWimFileRegistry : Failed while loading the registry hive

After that, Bootwiz crashes and leaves a DISM mount which must then be manually unmounted.

This is happening on all site servers, including a new device spun up specifically to test this. All are running Windows Server 2016, 2019, or 2022.

Our Altiris server is currently at 8.6 RU3 with no way to upgrade to 8.7 due to limitations around server OS upgrades (we're stuck on WS 2012 R2 because of a faulty component which effectively prevents upgrades, and Microsoft themselves have admitted there's no fix).

The WinPE optional components are enabled and sequenced correctly. I've also readded the Win10 1903 ADK + WinPE to NBS, and I've tried repairing the Deployment Solution component from within SIM. The BDC package has definitely been updated and redistributed to site servers for use, but the problem remains.

I have seen reference to a similar issue under 8.5 RU3 but, of course, the original TECHDOCS link is broken, and the wolkenservicedesk link sends my browser into a redirect loop.

I'll add the full Bootwiz log in the next message. Happy for any pointers or recommendations, or indeed, an actual fix.

Thanks

------------------------------
Tech Monkey/IT Primate
------------------------------

[deemacgee]

Bootwiz log file.

------------------------------
Tech Monkey/IT Primate
------------------------------

[deemacgee]

After a bit more keyboard slamming, I've been able to get a new PXE image to successfuly compile after deleting the entire waik_winpe10 directory from ..\Deployment\BDC on the NS itself, reloading the ADK+WinPE source into NBS, and then updating the BDC package for distribution to site servers. The old preboot still fails.

However, for reasons unknown, I still have to delete the any instance of the current BCD package from the ..\Deployment\Downloads directory on each site server before the new image compiles properly. It's not ideal, but I'll take it.

Hopefully this thread will save other people the same hassle.

------------------------------
Tech Monkey/IT Primate
------------------------------

[deemacgee]

Another update - Bootwiz has started crashing again when generating the second of our new WinPE images. Preboot creation happens on a loop on each server and crashes with the same error log as posted above. In some cases it is stuck for several hours; in others the bootwiz process is idle and must be manually terminated.

Our Linux PE boot is not affected.

Does anyone at Broadcom know how to completely reset the NBS component to baseline, and whether this will break our existing jobs and task sequences? I'm not bothered about complexity or micromanagement or time, I just need to get NBS working again.

------------------------------
Tech Monkey/IT Primate
------------------------------

[deemacgee]

Hello everyone, resurrecting this thread because we're still having significant issues compiling WinPE boot images.

Our latest failure is a Windows Server 2022 device, everything freshly installed, no legacy anything, Altiris 8.6 RU3. NBS installs without issue. The Windows ADK on the Altiris server was Windows 10 1909 (10.1.18362.1); as part of today's diagnostic work, it was updated to Windows 10 2004. BDC package is replicated to the server and has been verified by package integrity check.

Bootwiz.exe process launches, mounts the WIM, and begins the process of boot image creation.

For reasons which aren't clear, the process crashes and a broken mount is left at C:\Windows\Temp\~bootwiz\[random]\image\, which then needs to be manually cleared. Each subsequent attempt to compile the image results in the creation of another directory in that path with a different [random] value. Unless dismounted, repeated attempts to compile the image will eventually fill the drive.

Logs are exactly as the one posted above, with the last entry reading:
>CPlatformWinPe3:UpdateWimFileRegistry : Failed while loading the registry hive

There is no other information beyond this point. No other logs are written, no useful information is available in any other NBS log files, and there is no indication of the cause of failure. The log file up to the point of failure appears to be normal, with no obvious errors visible.

We have dealt with this issue intermittently over the life of ITMS and have simply retried compiling boot images until successful (2, 3, sometimes 5 attempts), but in the last ~9 months, we've had only failures. We're also unable to successfully recompile any of our existing boot images on other NBS servers (running Windows Server 2016/2019/2022); as a result, we cannot update drivers or add data/utilities to the boot image without manually updating each WIM on 150+ PXE servers. The platform is rapidly becoming unfit for purpose, and I expect the issue to become critical with the next generation of hardware.

I can find no references to this issue anywhere.
Is this behaviour known at all to anyone at Broadcom or anyone in the user community?

Thanks

------------------------------
Tech Monkey/IT Primate
------------------------------

[Martin C. M. Dhiin]

Hi,

Have you by any chance created a support case on this issue?

Based on all the error messages in the previous log file, loads of errors appear:
"returning thread exitcode 600 (BDC_ERR_WIM_UNMOUNT)"
SetNamedSecurityInfo call failed: 2
WIMUnmountImage failed - GetLastError returned -1052638948 ()

Some of these error messages could be related to permissions from the service account that could be corrupted somehow, or an antivirus exclusion not being made to C:\Windows\Temp\ or C:\Windows\Temp\~bootwiz.

Depending on your setup is it a dedicated service account you are using? If it is a dedicated service account try changing it to a build in service account and see what happens. If same result change to the dedicated service account.