TL;DR: All is now well.
On the second server I was able to remove the old agent, reboot it, install the new agent, successfully conduct a rescan which put it into a weird management status of sorts, rebooted it a second time which upon logging in spat out an error from Veeam saying it couldn't do something because it was being centrally managed. I then did another rescan which then said it had to repair the agent install... but when it finished it said it "failed." I suspect that it was because it also said the agent/protected server needed to be rebooted... which I did.
When it came back up from this 3rd reboot there were no error messages upon login this time. I ran yet another rescan of the server and this time it passed with flying colors. I then tested taking a backup of the server and that failed. To no surprise, the backup job complained that the Change Block Tracking map was changed for the volumes on the server. But the failure wasn't about CBT... it was an authority error... "No authority could be contacted for authentication." I've seen this one before and was reported to have been an issue with contacting the Domain Controller (of which there are 2 online). So... I just kicked it off again to see what may happen. This time "Job has failed unexpectedly... Managed session <jibberish string> has failed." So... I rebooted both the agent/protected server and the VBR server.
After both were rebooted, just for good measure I ran a rescan of the agent server which was successful. I then once again kicked off the backup job. It once again spat out that CBT map for the volumes changed... but started processing much like before. This time however... It finished. An hour and a half later having read through 4TB of data and transferring less than 3 GB of backup data (which I have to admit was comical to see the compression reported as ">999x"), the job had finished with warnings about the CBT map.
Now, with a CBT map change error it will often times clear on the subsequent run. But, I've also had a standalone agent give this error repeatedly to no end which was only resolvable by removing the Veeam CBT driver. As this is a managed server, I know how to manually remove it, but I'm not clear how to tell VBR to ignore auto-adding the CBT driver back on the server on the subsequent rescan. So, I crossed my fingers hoping for a successful followup job and kicked it off again. This time... Success! No CBT warnings and the job finished in under 5 minutes. Order has been restored!
So... 4 reboots of the agent server this time instead of just 2, and only 1 reboot of the VBR server. That was a bit painful, but ultimately (thankfully) came out the other side of it alive. Time to close out my support ticket.
