I considering using MongoDB change stream feature to replicate some data from one place to another. However, I am trying to figure out of I 1) first need to do a manual one-off copy (backfill), and then 2) initialise the change stream replication - or if the change stream simply will contain all data stored in MongoDB and automatically would copy all relevant data (that is, would imply backfilling).
The change stream uses the oplog. Since MongoDB 4.4, it can be configured to keep the oplog for at least
That is, one cannot use MongoDB change stream for backpropagation. Instead, backpropagation needs to happen through paginating through the actual data using the fetch API.