Background
Sharded mongo cluster with 5 shards
3 stand alone config server instances
Mongo version : 3.0.8
Requirement
Upgrading to 3.2.8 as well as converting individual config servers to replica set
Reference
Config servers to replica set conversion
Preparation
Tested successfully in QA environment.
Problem
The production scale is way more than QA and it is serving live traffic. Is there any way to have a roll back plan if the conversion from individual servers to replica set fails.
One possible solution can be : To back up the config server data and restore in case of failure, then restart 3 servers and mongos as well. How feasible is this ?