I see some of these in the nmm log
The Exchange backup is promoted to Full because the WriteTracker driver re-initialized. The driver might re-initialize because the system was rebooted or because database files were moved to a different volume.
The server has not been rebooted and the databases has not moved to a different volume.
I tried forcing an incr but it doesn't seem that nothing actually happens. I don't see above log mesage that the backup is being promoted to full in the log but the backup fails after a long time on the large mailbox databases.
Could it be that Exchange 2010 VSS writers just not handles the BBB as well as perhaps Exchange 2016 and this is "expected" behavior?