gRPC C++ client failover to other node listed in ipv4: name resolver.

[Eddie Mon]

Hello,

I am trying to figure out if there is a way for the gRPC server to trigger the gRPC client to retry a request on the next server node in the ipv4: address list without resorting to shutting down the server or the channel with the gRPC client.  

I have tried many different return codes, but it seems that the only way to make the gRPC client move on to the next server IP is by terminating the gRPC channel.

Does anyone know if what I want to do is achievable?

Eddie.

[Eddie Mon]

They are much appreciated if anyone has any insight to share. 

Thanks,
Eddie.

[sanjay...@google.com]

Pls check this out https://github.com/grpc/proposal/blob/master/A6-client-retries.md#overview specifically

We also provide a mechanism for servers to explicitly signal clients to retry after a settable delay. See detailed description of server pushback. With the correct LB policy the client could retry on a different IP address.

[Eddie Mon]

Thank You. I will look into that.