However, in our power cycle tests, we do the following:
1. Start a single node replica set on port 20000
2. Restart the node on port 20001
3. Node is unable to find itself in the config on startup, enters REMOVED
- Node does not start data replication and does not set the follower mode to RECOVERY
4. A reconfig is performed to update the hostname in the rsConfig.
5. Node finds itself in the config, transitions to STARTUP2, and has no means to get out.