Forum Discussion
Re: 2025-10 Cumulative Update for Windows Server 2019 (KB5066586) Undoes Update on Reboot
Hello,
Start by stopping non-Microsoft services with MSCONFIG.
Stop and disable SQL Server-type services.
Check if you have any unknown profiles in "Advanced system settings" -> User Profiles -> Settings.
If you have any, delete them.
Reboot your server.
Try installing your updates again.
That's it.
6 Replies
I performed these step.
- No SQL Server, so did not need to stop those services.
- I didn't have any unknown profiles, but I did find a bunch of profiles in there, as if people were logging into the server, even though they weren't. I deleted them all.
- MSCONFIG not running. All of the other services looked normal.
Unfortunately, the update did not install. By the way, it is now trying to install 2025-11 Cumulative Update for Windows Server 2919. This is what happens on the reboot.
Sorry for the slow respones. I was on a work trip.
I appreciate these tips. These are some good ideas. I'll give them a try this coming Saturday. I have to wait until the weekend, because this is a production server. I'll report back on Monday, Nov. 17.
I manually downloaded the 2025-12 Cumulative Update for Windows Server 2019 from the Windows Update Catalog. It did not install. I also tried an older 2025-07 Cumulative Update and it didn't install either.
Based on the link to the Microsoft Learn page, and reviewing the cbs.log file, it appears the drivers are not loading. There are a number of drivers that are failing to install. I'm not sure how to get around this.
As the "0x8007000d (ERROR_INVALID_DATA)" error message indicates a possible system corruption, please try below steps to repair the image, if machine is connected to the internet (DISM works best when it can connect to Microsoft Update to grab any missing or corrupted files):
- Open an elevated command prompt.
- Type the following command, and then press Enter. It may take several minutes for the command operation to be completed: DISM.exe /Online /Cleanup-Image /RestoreHealth
- Reboot machine and check if same error.
Otherwise, if your computer can't connect to Windows Update, you can alternatively use a working Windows installation as the repair source.
If still same issue, as a last resort, you could try identifying the invalid data and attempt to restore the files from similar Windows image in a healthy state, following guidelines in https://learn.microsoft.com/en-us/troubleshoot/windows-server/installing-updates-features-roles/error-invalid-data-startup-installing-updates
But, keep in mind that this process may be time-consuming and you might end up needing to rebuild the image after all (as described in article resolution).
I ran DISM.exe /Online /Cleanup-Image /RestoreHealth.
I went to the cbs.log file, and found this afterwards. Not sure if it is related to the update uninstalling itself on reboot.
2025-11-22 09:31:33, Info CBS All CSI corruption was fixed, ensure CorruptionDetectedDuringAcr is clear
2025-11-22 09:31:33, Info CBS Failed to clear CorruptionDetectedDuringAcr store corrupt flag (slow mode trigger). [HRESULT = 0x80070002 - ERROR_FILE_NOT_FOUND]
2025-11-22 09:31:33, Info CBS CheckSur: hrStatus: 0x0 [S_OK], download Result: 0x0 [S_OK]
2025-11-22 09:31:33, Info CBS Count of times corruption detected: 0
2025-11-22 09:31:33, Info CBS Seconds between initial corruption detections: -1
2025-11-22 09:31:33, Info CBS Seconds between corruption and repair: -1
2025-11-22 09:31:33, Info CBS Reboot mark cleared
During installation,
...and the same failure message.
I made a backup copy of the CBS log in case there is some useful information in there. I didn't find any failure messages.
