Client Error: io.grpc.StatusRuntimeException: UNKNOWN

What version of gRPC are you using?

1.0.2

What JVM are you using (java -version)?

1.8

What did you do?

I am running a gRPC-java server with multiple grpc services defined.

When running multiple stubs and their multiple rpc calls simultaneously I am seeing this exception below thrown.

io.grpc.StatusRuntimeException: UNKNOWN
	at io.grpc.Status.asRuntimeException(Status.java:545)
	at io.grpc.stub.ClientCalls$StreamObserverToCallListenerAdapter.onClose(ClientCalls.java:395)
	at io.grpc.internal.ClientCallImpl$ClientStreamListenerImpl.close(ClientCallImpl.java:481)
	at io.grpc.internal.ClientCallImpl$ClientStreamListenerImpl.access$600(ClientCallImpl.java:398)
	at io.grpc.internal.ClientCallImpl$ClientStreamListenerImpl$1StreamClosed.runInContext(ClientCallImpl.java:513)
	at io.grpc.internal.ContextRunnable.run(ContextRunnable.java:52)
	at io.grpc.internal.SerializingExecutor$TaskRunner.run(SerializingExecutor.java:154)
	at java.util.concurrent.ThreadPoolExecutor.runWorker(ThreadPoolExecutor.java:1142)
	at java.util.concurrent.ThreadPoolExecutor$Worker.run(ThreadPoolExecutor.java:617)
	at java.lang.Thread.run(Thread.java:745)

This happens only when I have multiple stubs with their own channel defined and the occurrence of the exception seems to be more frequent when increasing the number of stubs/rpc calls. This looks to me like some sort of resource issue either client or server side. I have increased the thread count and seen no measurable improvements. Not sure where to go from here?

My server setup is as follows:

private final NettyServerBuilder builder;
private final NioEventLoopGroup workerGroup;
...
...

// The options are currently set to the defualt grpc settings. So in effect these should not change the default environment.
// NioEventLoopGroup( 2 * Runtime.getRuntime().availableProcessors() )
builder.workerEventLoopGroup( workerGroup )
                              // 1048576
				.flowControlWindow( options.getFlowControlWindow() ) 
                              // Integer.MAX_VALUE
				.maxConcurrentCallsPerConnection( options.getMaxConcurrentCallsPerConnection() )
                              // 8192
				.maxHeaderListSize( options.getMaxHeaderListSize() )
                              // 4 * 1024 * 1024
				.maxMessageSize( options.getMaxMessageSize() )
                              //ProtocolNegotiators.serverPlaintext()
				.protocolNegotiator( options.getProtocolNegotiator() ) 
                              // Executors.newFixedThreadPool(20)
				.executor( executor )
                              // CompressorRegistry.getDefaultInstance()
				.compressorRegistry( options.getCompressorRegistry() )
                               // DecompressorRegistry.getDefaultInstance()
				.decompressorRegistry( options.getDecompressorRegistry() );

My Client setup:

ManagedChannel channel = NettyChannelBuilder.forAddress( remote_host, remote_port )
				.usePlaintext( true )
				.build();

Runtime.getRuntime().addShutdownHook( new Thread( new Runnable() {
			@Override
			public void run() {
				channel.shutdown();
				try {
					channel.awaitTermination( 120, TimeUnit.SECONDS );
				} catch ( InterruptedException e ) {
					Thread.currentThread().interrupt();
				}
			}
		} ) );

CallOptions callOptions = CallOptions.DEFAULT.withCompression( "gzip" );
// Where stubClazz would be the grpc-proto stub
Constructor<Stub> constructor = stubClazz.getDeclaredConstructor( Channel.class,
				CallOptions.class );
constructor.setAccessible( true );
return constructor.newInstance( channel, callOptions );

What did you expect to see?

I expect to get back the response object.

What did you see instead?

Instead I received a io.grpc.StatusRuntimeException: UNKNOWN

[ejona86]

The server probably threw an exception. The exception information is not safe to propagate to the client by default, so you get a useless error on client-side. Did you check the server logs to see if there were any uncaught exceptions?

[wrightm]

@ejona86 apologies typo, that sorts out the unsupported message type: Object (expected: ByteBuf, FileRegion) exception. Thank you.

And also looks like it was the server throwing an exception like you said. However, grpc is not logging this to my file appender with my current logging configuration. Using log4J, I can now look into why this is the case.

Good old sql util Date conflict… There isn’t a Date.toInstant in sql.Date just a java.lang.UnsupportedOperationException, there was me thinking java 8 sorted things like this out. I guess it make sense since sql.Date only support date, month, and year.