Destination clusters in Atlas should have:
⬚ Adequate storage, IOPS, connections, and RAM. Note: the destination deployment needs to be LARGER because of index building during initial sync.
⬚ No traffic or workload. Cluster should be empty.
Source clusters should have:
⬚ Adequate oplog. Please read this and check oplog configuration.
⬚ The required network access
⬚ Network bandwidth. Please confirm corporate network will not limit/refuse the amount of data that will be transferred to Atlas.
⬚ Authentication enabled on source deployment (strongly suggested for the migration)
⬚ The correct user is being passed with correct roles. Please ensure source cluster has the user set up in admin database. Ensure you have granted the Atlas Admin role to the destination user as per docs here.
⬚ The proper version of MongoDB (at least 2.6)
⬚ If the destination cluster is a more recent version than the source cluster: driver must be compatible with the new version. See here.
⬚ The latest version of the MongoDB Shell
Notes:
Live Migration cannot utilize a VPC peering connection; if using a VPC peering connection for the migration is required, use `mongomirror` instead.
You cannot resume the process of migration, it must restart from beginning. In which case, migration might be best accomplished via scheduled time, rather than a time of peak traffic.