GRPC java 双向流式服务，直接内存溢出解决

客户端解决：

public class ManualFlowControlClient {
    private static final Logger logger =
        Logger.getLogger(ManualFlowControlClient.class.getName());

  public static void main(String[] args) throws InterruptedException {
    final CountDownLatch done = new CountDownLatch(1);

    // Create a channel and a stub
    ManagedChannel channel = ManagedChannelBuilder
        .forAddress("localhost", 50051)
        .usePlaintext()
        .build();
    StreamingGreeterGrpc.StreamingGreeterStub stub = StreamingGreeterGrpc.newStub(channel);

    // When using manual flow-control and back-pressure on the client, the ClientResponseObserver handles both
    // request and response streams.
    ClientResponseObserver<HelloRequest, HelloReply> clientResponseObserver =
        new ClientResponseObserver<HelloRequest, HelloReply>() {

          ClientCallStreamObserver<HelloRequest> requestStream;

          @Override
          public void beforeStart(final ClientCallStreamObserver<HelloRequest> requestStream) {
            this.requestStream = requestStream;
            // Set up manual flow control for the response stream. It feels backwards to configure the response
            // stream's flow control using the request stream's observer, but this is the way it is.
            requestStream.disableAutoInboundFlowControl();

            // Set up a back-pressure-aware producer for the request stream. The onReadyHandler will be invoked
            // when the consuming side has enough buffer space to receive more messages.
            //
            // Messages are serialized into a transport-specific transmit buffer. Depending on the size of this buffer,
            // MANY messages may be buffered, however, they haven't yet been sent to the server. The server must call
            // request() to pull a buffered message from the client.
            //
            // Note: the onReadyHandler's invocation is serialized on the same thread pool as the incoming
            // StreamObserver's onNext(), onError(), and onComplete() handlers. Blocking the onReadyHandler will prevent
            // additional messages from being processed by the incoming StreamObserver. The onReadyHandler must return
            // in a timely manor or else message processing throughput will suffer.
            requestStream.setOnReadyHandler(new Runnable() {
              // An iterator is used so we can pause and resume iteration of the request data.