Hi @kevinadi ,
Thank you for the suggestions! At the moment of the lag the activity on the replica when it comes to cpu/disk actually drops, so it looks more like a locking issue. But I don’t understand how it can be related to the distance. It’s a curious combination of factors.
We are on MongoDB 4.4. and streaming replication already. I have a thought to try to switch it to the batch mode to see if it gives any difference (to possibly prove it’s some streaming replication issue), but have not done this yet.