While fixing bug-3221, if leader is marked unreachable, the PrimaryShard Cache in actorcontext should also be invalidated.

A possible solution is to make actorcontext to listen to ShardInformation changes. If we do that, then we can possibly update the findShard code.
For any local shard changes, actorcontext will receive information via listener and in that case, we don't have to keep that in cache. For any remote Shards (if a shard replica does not exist on local) a FindShard message could be sent.