My coworker said I was crazy for doing an update near the end of the day, on a Friday, and also being Friday the 13th. I told him to relax. Upgrading Veeam to V13.1 is a breeze!! The good news is, it was, except for one small hiccup.
I recently completed an upgrade of our Veeam Backup & Replication environment to v13.1. The experience was straightforward and uneventful for the most part. The VBR server upgraded successfully, Enterprise Manager upgraded cleanly, repositories were upgraded, and most managed servers updated as I expected they would.
The only issue encountered involved a Windows proxy/repository server that repeatedly failed during component upgrade. This however became a slightly larger issue near the end of the install process.
Symptoms
The Veeam server and most components were all successful, so this was one of the last steps, the pop up was annoying, and this is a critical piece of infrastructure so I needed to get it going.
The affected server was a Windows Server proxy/repository host.
Every attempt to upgrade the components would fail with:
Failed to upgrade host components.
Error 1920. Service Veeam vPower NFS Service (VeeamNFSSvc)
failed to start. Verify that you have sufficient privileges
to start system services.
At first glance this appears to be a permissions issue. Since the installer specifically references privileges, that was my initial assumption as well.
However, local administrator rights were confirmed and the server had no obvious issues.
Troubleshooting
The first few troubleshooting steps were fairly standard:
- Rebooted the server.
- Retried the component deployment.
- Verified no pending reboot state.
- Checked local administrator permissions.
- Reviewed installed Veeam components.
- Removed an old Veeam CDP Proxy component that had remained at an older version.
None of these changed the outcome.
The deployment continued to fail while installing the vPower NFS component.
I even totally uninstalled all Veeam programs, and would try to add them back but it would keep failing with the same error.
Logs
The breakthrough came from reviewing: (I reviewed ALL the logs I think and it took a while to find it)
C:\ProgramData\Veeam\Backup\Svc.VeeamNFS.log
The NFS service was actually starting and binding to its ports:
Running synch tcp server on port 1063
Running synch tcp server on port 2049
Running synch tcp server on port 111
Then immediately failing with:
Cannot register on portmapper.
NFS program mismatch. Low: 2, High: 2
Failed to start service.
The vPower NFS service was colliding with something that already owned the RPC portmapper functionality… At least that’s what a lot of searching and a few AI tools helped tell me as the error wasn’t very clear.
Conflict
The next step was determining what owned port 111.
netstat -ano | findstr ":111"
Result:
TCP 0.0.0.0:111 LISTENING 3740
UDP 0.0.0.0:111 3740
Then identifying the process:
tasklist /fi "PID eq 3740"
Result:
portmap.exe
Finally mapping the process back to a service:
Get-CimInstance Win32_Service |
Where-Object {$_.ProcessId -eq 3740} |
Select Name,DisplayName,State,StartName
Result:
Name : oncportmap
DisplayName : ONC/RPC Portmapper
What the heck is this? Who installed it? after a bit more digging, the service executable was QLogic. I knew right away that that was part of their driver install for the Fiber Channel adapters on the proxy.
C:\Program Files\QLogic Corporation\PortMap\portmap.exe
Cause
The server still had legacy QLogic Management Suite components installed from years ago. It likely was installed with the drivers.
The associated ONC/RPC Portmapper service was listening on TCP/UDP port 111 and was conflicting with the vPower NFS service registration process.
This explained the log entries:
Cannot register on portmapper.
NFS program mismatch.
Failed to start service.
and also explained why the installer's Error 1920 message was misleading.
The service wasn't failing because of permissions.
It was failing because it could not successfully register the required NFS services with the existing portmapper implementation.
The Veeam install was perhaps thinking it didn’t have the proper permissions to uninstall the previous or new service, but at least the log files had some good information.
Resolution
The following legacy components were removed: Nothing fancy, Add remove programs, uninstall the Qlogic Managment Suite and Portmapper. I had the options to leave the drivers which are still in use.
QLogic Management Suite Java Agent
QLogic Management Suite Java iQAgent
QLogic Management Suite Java nQLRemote
ONC/RPC Portmapper
After uninstalling the software I wanted to verify the ports were free to not waste my time running the infrastructure upgrades over and over again.
netstat -ano | findstr ":111"
returned no listeners.
I immediately retried the Veeam component deployment.
The upgrade completed successfully and the vPower NFS component installed without issue.
No reboot was even required after removing the QLogic software.
Lessons Learned
A few takeaways from this experience:
- Error 1920 is not always a permissions issue.
- The Veeam service logs provided a much more accurate explanation than the installer error.
- If vPower NFS installation fails, check:
Svc.VeeamNFS.log- TCP/UDP port 111
- RPC portmapper services
- Legacy storage-management software may introduce hidden dependencies that don't become apparent until a major upgrade.
- If you still have old QLogic Management Suite software on proxies or repositories, review whether it is actually required before your next upgrade window.
Final Thoughts
Aside from this issue, the Veeam 13.1 upgrade was very smooth and uneventful. The backup server, Enterprise Manager, hardened repositories, and remaining infrastructure all upgraded successfully.
The only obstacle turned out to be a legacy ONC/RPC Portmapper installed with older QLogic management tools. Once removed, the managed server upgraded immediately.
Hopefully this helps someone else who runs into a similar vPower NFS failure during a Veeam 13.x upgrade and saves a few hours of log hunting.
Also, I feel like there is some type of lesson learned about production upgrades and updates on Fridays, at the end of the day, or long weekends. Don’t ask me though as I didn’t learn it. 🤣
