Since I’ve been working almost 24/7 over the holidays while migrating one huge environment (15.000 users) into a new one in a hurry. That is, without any previous knowledge of the old environment and a very tight schedule, I’ve been using all the shortcuts I can come up with. I quickly realised that PrintBRM would save my a*s when it came to migrating printers. I still ended up adding them more or less manually since I migrated from Windows Server 2003 x86 to a Windows Server 2008 x64-based cluster and the time finding the right drivers for it to work flawlessly would’ve been longer than the manual add.
Anyway, once I’ve added all my printers to my new cluster I came to realize that I had screwed up in the cluster configuration so I needed to backup my printers and rearrange a little (I DON’T want to add those printers manually again). Along came PrintBRM.exe which I used to backup the old Windows Server 2003 server with from a Vista machine so I gladly fired it up on one of my cluster nodes. It came about 58% before it terminated with the good old error message “The following error occurred: 0x80070043. The network name cannot be found.” while backing up a driver. I removed the failing driver and tried again, 60% this time. I realized that I’d be deleting printers and drivers all night before it would finish so I needed a workaround.
I did some googling and came up with more people with similar problems:
And the usual how-to’s from Microsoft:
And if you bother to read any of the above you’ll quickly notice that there’s no solution. So I’ll give you my workaround:
I looked with process monitor since it was mentioned in the first forum post above, the error thrown is “Bad network name”. Let’s say our print server is called “PrintSrv”, and the syntax of PrintBRM is:
This will cause PrintBRM to go looking for PrintSrvc$WindowsSystem 32 to get the dll’s for certain drivers. A clustered print server doesn’t have a c$ since it only has a spooler, hence it’ll fail.
Check your Failover Cluster Manager for which disk is your print server storage (in the picture below it’s O:)
The following steps need to be done on the node that has the resources for obvious reasons:
Open Explorer and share your drive (regardless of letter, remember my O:) as C$
Open the drive in Explorer and create a directory called “Windows” (yes, that’s correct)
The below information is incorrect, I’ve now found the mklink.exe application included in Windows Server 2008 which will do the same as linkd.exe.
The syntax of the command is: mklink /D O:Windows C:Windows
Then you’ll need to download the Windows Server 2003 Resource Kit since you’ll be needing the linkd.exe application. This application creates junction points which in our case is just what we need.
If we assume that O: is our drive the syntax of linkd.exe will be: linkd O:Windows C:Windows
This will create a junction point (you can see this in explorer both on the arrow and the description) from O:Windows to C:Windows.
That’s it! Now you can backup all your printers regardless of driver.
Why it works? Well, we’ve shared our drive as c$ so now your clustered print server actually has a C$, and the junction point gives it ServerNameC$Windows and all the files PrintBRM is looking for are located there.
When you’re done you’ll need to remove the junction point: linkd O:Windows /D (D = Delete) and don’t forget to remove the sharing of your drive as C$ either.
This is filed as a bug with Microsoft so I’m hoping for a QFE / patch as soon as possible.