[grpc-java] NameResolver changes in 1.1.1 ?

[Jorg Heymans]

Hi,

After updating to grpc 1.1.1 it seems that the semantics around using NameResolver have changed. Basically in our (working) 1.0.3 NameResolver implementation we were doing this in start() :

      Collection<ServiceInstance> serviceInstances = provider.getAllInstances(); // provider is a curator.ServiceProvider

      List<ResolvedServerInfo> infoGroups = new ArrayList<>();

      for (ServiceInstance serviceInstance : serviceInstances) {

        infoGroups.add(new ResolvedServerInfo(

            InetSocketAddress.createUnresolved(serviceInstance.getAddress(), serviceInstance.getPort()),

            Attributes.EMPTY));

      }

      listener.onUpdate(Collections.singletonList(infoGroups), Attributes.EMPTY);

For 1.1.1 we had to slightly change it to match the listener.onUpdate signature change:

      Collection<ServiceInstance> serviceInstances = provider.getAllInstances();

      List<ResolvedServerInfo> serverInfos = new ArrayList<>();

      for (ServiceInstance serviceInstance : serviceInstances) {

        serverInfos.add(new ResolvedServerInfo(

            InetSocketAddress.createUnresolved(serviceInstance.getAddress(), serviceInstance.getPort())));

      }

      listener.onUpdate(Collections.singletonList(ResolvedServerInfoGroup.builder().addAll(serverInfos).build()),

          Attributes.EMPTY);

However when our client now tries to connect to the server we get an exception indicating the channel is not active. 

io.grpc.StatusRuntimeException: UNAVAILABLE: Channel in TRANSIENT_FAILURE state

at io.grpc.stub.ClientCalls.toStatusRuntimeException(ClientCalls.java:227)

at io.grpc.stub.ClientCalls.getUnchecked(ClientCalls.java:208)

Any thoughts what could be going on here ?

Thanks,

Jorg

[Kun Zhang]

We should have mentioned the signature change in our release note. Sorry about that.

The error doesn't seem to be related to the signature change. The message is logged from here, which indicates that the connect attempt has failed. If you turn on FINE logging, you may find something useful.

[Jorg Heymans]

Indeed the logging revealed in fact an unrelated error during SSL init, thanks for the hint.

Jorg