During logical initial sync, oplog events are applied on the new node and trigger OpObserver events as they are applied, which ensures that the updates are propagated to in-memory threads that consume these values. File-copy based initial sync, however, does not trigger the OpObserver hooks. This means that replica set nodes that are added via file-copy based initial sync will not have the correct values in-memory at the end of initial sync.
We should adopt an approach similar to SERVER-64601 in order to ensure that cluster parameters are properly initialized in-memory when file-copy based initial sync is used.
- is related to
-
SERVER-64601 Shard server mongod initialization is not run after File Copy Based Initial Sync
- Closed
-
SERVER-69015 Add cluster parameter consistency checks to ShardingTest
- Blocked
- related to
-
SERVER-64627 Need general method to handle in-memory state after initial sync
- Closed