1) Trigger a failpoint in migration before issuing the moveChunk.
2) Restart the shard with recoverShardingState=false (this setParameter only affects if the minOpTimeRecovery document is used to recover sharding state) and without --shardsvr.
3) Verify that minOpTimeUpdaters in the minOpTimeRecovery document is non-zero.
3) Restart the shard with recoverShardingState=true and without --shardsvr.
4) Verify that minOpTimeUpdaters in the minOpTimeRecovery document goes to zero.
5) Verify that the (config)opTime in the recovery document has been updated.