The issue as I understand it is booting to a partition larger than 2TB, not necessarily having one. If this is to be a file server, then the goal IMO should be to have a maximum sized single storage partition. There is no valid reason why the users would need access to the file servers OS files.
Put the OS on a separate 100GB partition. Then copy over the old 2GB partition, convert it to GPT, and resize it to take up the rest of the remaining space.
When applying 3TByte disks in my computers I always had to use software from Seagate that enabled my computers to make use of storage capacity above 2TByte and always ended up with 2 (GPT) partitions in the process (one of 2TByte and one with the remaining storage space). Happened even on a Asus motherboard that I bought less than a year ago.
I remember Seagate's DDO software from years ago (Win2k days), when the IIRC 8 GB limit was all the rage. It had three very bad habits then...that I would still suspect it of pulling now:
It never verified if it's presence was actually necessary, it just auto installed.
It made it impossible for Win2k to boot, because it was just in the way.
It was next to impossible to find and remove - I basically had to beat it to death with a hammer while suffering through a ton of dire and ominous "warnings" before it finally released its demonic hold on the drive. Now after
the exorcism the drive - which previously couldn't make it through the 2k install without BSODing - ran Win2k perfectly for several years.