Custom hadoop rpc plugins

[yuyanlei-8130]

To monitor the links between hadoop clients and hadoop servers, I refer to https://skyapm.github.io/document-cn-translation-of-skywalking/zh/6.3.0/guides/Java-Plugin-Development-Guide.html,It seems that hadoop can't implement such a plugin, and the hadoop rpc is issued like this:

public void sendRpcRequest(final Call call) throws InterruptedException, IOException {
      if (shouldCloseConnection.get()) {
        return;
      }

      RpcRequestHeaderProto header = ProtoUtil.makeRpcRequestHeader(
          call.rpcKind, OperationProto.RPC_FINAL_PACKET, call.id, call.retry,
          clientId, call.alignmentContext);

      final ResponseBuffer buf = new ResponseBuffer();
      header.writeDelimitedTo(buf);
      RpcWritable.wrap(call.rpcRequest).writeTo(buf);

      synchronized (sendRpcRequestLock) {
        Future<?> senderFuture = sendParamsExecutor.submit(new Runnable() {
          @Override
          public void run() {
            try {
              synchronized (ipcStreams.out) {
                if (shouldCloseConnection.get()) {
                  return;
                }
                if (LOG.isDebugEnabled()) {
                  LOG.debug(getName() + " sending #" + call.id
                      + " " + call.rpcRequest);
                }
                // RpcRequestHeader + RpcRequest
                ipcStreams.sendRequest(buf.toByteArray());
                ipcStreams.flush();
              }
            } catch (IOException e) {
              // exception at this point would leave the connection in an
              // unrecoverable state (eg half a call left on the wire).
              // So, close the connection, killing any outstanding calls
              markClosed(e);
            } finally {
              //the buffer is just an in-memory buffer, but it is still polite to
              // close early
              IOUtils.closeStream(buf);
            }
          }
        });
      
        try {
          senderFuture.get();
        } catch (ExecutionException e) {
          Throwable cause = e.getCause();
          
          // cause should only be a RuntimeException as the Runnable above
          // catches IOException
          if (cause instanceof RuntimeException) {
            throw (RuntimeException) cause;
          } else {
            throw new RuntimeException("unexpected checked exception", cause);
          }
        }
      }
    }

proto is an rpc request sent asynchronously. In this case, how can I develop a plugin to monitor the rpc between hadoop client and hadoop server?

[wu-sheng]

If it is only a Hadoop client, I don't think you need to add headers. This plugin should work like a MySQL client, right?

[yuyanlei-8130]

Is there a hadoop plugin? I couldn't find one

[wu-sheng]

No, and that would be not easy work to trace executions inside the Hadoop cluster.
Why would you like to do this?

In your question context, if an RPC doesn't have a header/metadata to add custom things, you have to change the body of the request.
This would make your remote peer sensitive to the new protocol. This is inevitable

[yuyanlei-8130]

I wanted a quick fix for locating hadoop
Like I mentioned with hadoop rpc, I have to modify the hadoop source code to adapt to it?

[wu-sheng]

You need to be clear about your question, what do you mean can't, which part do you mean can't?

[wu-sheng]

async RPC call is not an issue. There are many aync RPC framework supported. And proto is only about serialization, not RPC.
So, I can't follow up the whole thing. I thought you mean header can't be injected.