In-Place upgrade Windows 2012 R2 to Windows 2016 on a DL380 Gen9 server

[chillia]

Like many of us,  I'm trying to beat the October 10, 2023 end of support date for Windows 2012.  We have a few physical servers running 2012 R2 that we have little choice but to do an in-place upgrade to Windows 2016. I know it's not recommended or preferred but there's no funds for new servers.

I've in-place upgraded two DL380 Gen9's from 2012 R2 to 2016 by mounting the 2016 .iso, and running setup (keeping all personal files & setting, etc.) The upgrade went flawlessly. Then I updated all the drivers and firmware using the latest SPP.  The server seems happy, The applications all test out and are running as expected.  My customer seems happy! The only issue is that when we go to install the latest Windows cumulative update, in this case the 2023-08 update, the update gets all the way to 100% and then says "We couldn't complete the updates, Undoing changes, Don't turn off your computer.

I've followed every Windows Update repair solution I can find.

 DISM /Online /Cleanup-Image /ScanHealth - (No file corruptions are detected)  

 DISM /Online /Cleanup-Image /RestoreHealth - (No file corruptions are detected) 

 SFC /ScanNow - (No file corruptions are detected) 

 net stop wuauserv

 net stop cryptSvc

 net stop bits

 net stop msiserver

 ren C:\Windows\SoftwareDistribution SoftwareDistribution.old

 ren C:\Windows\System32\catroot2 Catroot2.old

 net start wuauserv

 net start cryptSvc

 net start bits

 net start msiserver

None of these steps resolves my issue of getting windows updates to successfully apply. It doesn't matter if the update comes from WSUS, Microsoft online or downloading the package and manually installing it. I'm thinking there is a driver mismatch between the hardware and the new OS but I'm unable to determine what that is. The Windows update logs aren't much help at all.

I thought I'd reach out to this forum before opening a support ticket with either Microsoft or HPE.

Thanks

Chris Hilliard

[MV3]

Hello 

Upon referring to the HPE Servers Support & OS Certification Matrices, DL380 Gen9 supports below OS.

WS2012
WS2012 R2
WS2016
WS2019

Since you have upgraded the OS instead of installing from the scratch, We suspect it to be an OS issue.
Engage Microsoft for further analysis. Request them for a detailed analysis if they point towards any hardware\driver.

Cheers...

[chillia]

Thank you for the reply back.  Contacting MS will be my next step or do another in-place upgrade to 2019 provided the apps on the server will support 2019.

When installing a OS fresh on a server I know that Intelligent Provisioning injects a lot of drivers, etc. during the install process.  When doing an in-place upgrade that same injection process does not occurr so I thought it worth reaching out to HPE first in case there where possibly some drivers that might not get picked up during the upgrade.  I'm assuming applying the latest SPP would have resolved that though.

[MV3 · Accepted solution]

Yes, you can try updating the SPP. If the issue persists even after an update, Microsoft will have to help us with the analysis of the issue reported.

Cheers...

[Sunitha_Mod]

Hello Chris,

Let us know if your concern has been addressed.

If you have no further query and you are satisfied with the answer then kindly mark the topic as Solved so that it is helpful for all community members.

[rASeeberger]

 @chillia 

Hi Chris

Have you found a solution for your issue? I'm planning the exact same upgrade path and would love to hear if Windows Updates started working afterwards?

Thanks

Luca

[chillia]

I've still not been able to get my 2 servers that I in-place upgraded from 2012 R2 to 2016 to accept the cumulative windows updates.  It's the same issue where the update process gets to 100% and then says it can't complete and rolls everything back.  I was waiting for the 2023-09 update to see if the issue might have been related to the 2023-08 update but the behavior is the same. I have not opened a support request yet with Microsoft as it just might be easier to rebuild the servers than to  deal with MS.  One of the many things I read about windows updates not being able to complete is that the recovery partition is not big enough.  Oddly enough, each of these servers has 2 recovery partitions, one is 450 MB, the other is 13.97GB.  The 13.97 GB partition resides immediately after the 450MB paritition so if the update process is trying to enlarge the 450MB partition it will surely fail.  According to the  reagentc /info command, the WIN RE is pointing to the 450MB partition so it should be pretty save to delete the odd 2nd recovery partition and see if that makes a difference with the update process.  It's worth noting that  I've done an in-place upgrade on a 3rd server that does not have this odd 2nd recovery partition and it's accepting the cumulative updates just fine.  I'm not sure how the 2nd recovery parition got created in the first place but deleteing it makes a lot of sense to me. 

[rASeeberger]

Thank you for getting back to me! Interestingly, I didn't face the issue when I performed an inplace upgrade on my Gen9 this past weekend from 2012r2. I was careful to install the Firmware and Driver using the HP SPP Package BEFORE proceeding with the upgrade. Instead of booting from it, I had to mount the .iso and initiate the SPP from the OS. This ensured that the server commenced driver installations rather than just focusing on Firmware. Post-upgrade, I did run into some problems with NW teaming. The inplace upgrade disrupted the teaming and prevented me from using the same pair of NICs to reestablish it. This might have been because the configuration was still partially lingering in the registry. Rather than troubleshooting, I simply used ports 3/4, the two other available ports, to set up a new teaming, and that seemed to do the trick.

Fortunately, the Windows Updates post-upgrade went on smoothly without any hitches.