gRPC has a native support for health-checking remote services as per https://github.com/grpc/grpc/blob/master/doc/health-checking.md

In our setup we've seen situation when Namerd becoming "broken" occasionally due to reasons not identified yet. In such situations it doesn't accept any new connections and all connected Linkerd instances are not able to resolve new names anymore. This is a separate issue, but having Linkerd health-checking Namerd and re-connecting to another healthy instance would've increased the stability in our case.