File-based initial sync initializes a MongoDB replica-set member by copying database files instead of logically copying documents. MongoDB 6.0 and later provides it as a MongoDB Enterprise method intended for very large datasets, typically 1 TB or more. section 9.1.3
Use the approach with these operational considerations:
- Choose it for large data sizes where logical initial sync may take too long. section 9.1.3
- Treat recovery windows differently: MongoDB can resume a file-based initial sync within a shorter 10-minute window after a failure, compared with 24 hours for logical initial sync. section 9.1.3
- Configure the relevant retry windows with mongod parameters, and consider oplog retention when preventing entries from being trimmed too soon. section 9.1.3
Large-member initialization
For a replica set containing more than 1 TB of data, a MongoDB Enterprise deployment can use file-based initial sync rather than copying every document logically. If a transient network problem interrupts the process, the shorter configured recovery window becomes an operational constraint. section 9.1.3
