To clarify how it works
We have a dns record ex. primary.mongodb.example.com this should always resolve to the ip of mongodb replica set primary node

On all replica set nodes there is a custom script that runs every minute  that connects to mongodb and checks which node is the primary

If the primary node changes than this scripts connects to the dns server and updates  the ip to which this primary.mongodb.example.com points to

So in jcon.config you only put this url primary.mongodb.example.com and the infrastructure makes sure that when you resolve this url you will get the current primary node ip

Regarding

"If the node mentioned in that file goes down then how automatically other node will work without any disruption in the service" there will be a short distruption of service consisting of 2 things

1. Time that is needed for mongodb to elect a new node (it can take from couple of seconds to 20s)

2. Time the script takes to find out that a new primary node is elected and to update the dns (script execution time is under 2s but it's run every minute so if the primary node changes ex. 10:10:30 the script will pick it up at 10:11:00 )
worst case scenario there will be a 90s disruption in mongodb availability

"If updated somehow it will require a server restart and which will have its own downtime"

I can either create a script that automatically restarts CHAT service if i see errors (ex. mongodb connection failure )  in the logs or the app itself if it fails to connect to mongodb can automatically retry each time resolving the mongodb url from config.json