Mongo replicaset init sync issue
When we have large size mongo replica set, we may have to take out each nodes for maintenance for few hrs or days.
In that case we have to increase size of Oplog to keep several days of transactions to resync nodes after maintanance.
If the RS headroom falls below oplog window, then we have blow away data and do init sync to add node back after maintanence.
The problem with very large size Oplog, will slow down any Change Stream process. Also it will occupy un neccasary space within collections.
Can mongo offer alternative way to constantly dump oplog data to local file system? Then we can use this data to re-apply transactions, to sync failed nodes, instead of doing init sync. Some thing similar to archichive log mechanism for redo log, Oracle uses.
This will avoid init sync issue for very large replica sets.