I had a corrupt file on a 6 terra byte production database, because of which the database would not come back online. I brought up the database in emergency mode "Alter Database Set Emergency" and ran a dbcc checkfilegroup on the filegroup of the corrupt file. The advice was to run a checkdb with repair_allow_data_loss.
The checkdb command is now running for over 48 hours and I have no clue when it will be over. The database size is about 6 TB and has about 800 files (due to partitioning).
Please let me know if you can extrapolate (from your experience) an approximate ETA for this checkDB operation.