Interesting. Can you confirm if the IP points to alpha3 (or alpha running with -o 102)? The port 7082 is used for internal grpc communication.
@praneelrathore if you’re sure that there was a crash, can you attempt the same operation in your staging/local environment? Trying to reproduce a crash on prod might not be the best approach 
If a similar issue happens, you should definitely save the logs. If you have monitoring setup on dgraph, you should be actively monitoring the memory and CPU when the tablet rebalance happens. If you see any unusual activity on memory/CPU, you should take a cpu/memory profile so that we can find the root cause.
When tablet move happens, the predicate being moved is read-only. Were you trying to write to the same predicate?
You don’t need to anything to distribute the predicates. Dgraph zero will monitor the predicate sizes and it will move them automatically. See https://dgraph.io/docs/master/deploy/cluster-setup/#shard-rebalancing
Also, I highly recommend reading Dgraph Whitepapers: In-Depth Insights and Analysis if you’re interested in understanding dgraph internals.