There are certain cases where the MongoDB process isn't producing enough logs:
- If a MongoDB process crashes when an index build is in progress and the index build is retried (replica set member or standalone not using noIndexBuildRetry) the index build is performed before the MongoDB process initializes replication.
- No explicit log entries are made regarding the retried index build.
If a MongoDB process recovers the last clean checkpoint and there are many time consuming operations to rollback. This is also done before the MongoDB process initializes networking and replication.
Please add more explicit logging for this kind of recovery operations performed at mongod startup.