[jgroups-users] v3->v4: my ReceiverAdapter methods are not being called

[Questions/problems r. to u. J. <jav...@li...>]

Thank you for previous help with castMessage/handle and
serializable->streamable. That all went fine, and I can send messages that
way properly now across the cluster. Now I'm seeing a bigger problem, but
it's so easy to describe that am hopeful it's a simple RTFM thing (though
I've read it several times now).

None of my ReceiverAdapter methods are being called. Am trying 4.0.8 now
but also tried 4.0.7. I can tell from logs/debugger that nodes are finding
each other and joining, are sending messages from one to the other [1], and
a getState call is made [2]. I override receive, get/setState,
viewAccepted, and suspect, and none of them are called. Note that this same
code worked with v3.

My base class is declared with "extends ReceiverAdapter implements
RequestHandler" and I set up like this (unchanged from v3):

        jChannel = env.createJChannel(); // tried with name(<string>) also
        jChannel.setReceiver(this);
        dispatcher = new MessageDispatcher(jChannel, this);
        jChannel.connect(<name>, null, 10 * 1000);

With channel creation like this:

        Protocol[] stack = {
            new TCP()
                .setValue("bind_addr", InetAddress.getByName(<bind
address>))
                .setValue("bind_port", <port>)
                .setValue("thread_pool_min_threads", 1)
                .setValue("thread_pool_keep_alive_time", 5000)
                .setValue("send_buf_size", 640000)
                .setValue("sock_conn_timeout", 300)
                .setValue("recv_buf_size", 5000000),
            new TCPPING()
                .setValue("initial_hosts", <list>)
                .setValue("send_cache_on_join", true)
                .setValue("port_range", 0),
            new MERGE3()
                .setValue("min_interval", 10000)
                .setValue("max_interval", 30000),
            new FD_ALL()
                .setValue("timeout", <timeout>),
            new VERIFY_SUSPECT()
                .setValue("timeout", 1500),
            new BARRIER(),
            new NAKACK2()
                .setValue("use_mcast_xmit", false),
            new UNICAST3(),
            new STABLE()
                .setValue("desired_avg_gossip", 50000)
                .setValue("max_bytes", 4000000),
//            createAuthProtocol(), <-- commented for now
            new GMS()
                .setValue("join_timeout", 3000),
            new MFC()
                .setValue("max_credits", 2000000)
                .setValue("min_credits", 800000),
            new FRAG2(),
            new STATE_TRANSFER()};
        return new JChannel(stack);

Since *none* of my ReceiverAdapter code is being called, am hoping it's
something dumb. I stepped into the receive case to try to figure it out and
found here is where the message is being swallowed:

RequestCorrelator, called by MessageDispatcher. hdr = null:

    public void receiveMessageBatch(MessageBatch batch) {
        for(Message msg : batch) {
            Header hdr=msg.getHeader(this.corr_id);
            if(hdr == null || hdr.corrId != this.corr_id) // msg was sent
by a different request corr in the same stack
                continue;

            if(hdr instanceof MultiDestinationHeader) {
                [...]
            }
            dispatch(msg, hdr);
        }
    }

There is no header that matches this corr_id (200). The headers are
NakAckHeader2 (41) and
TpHeader (57).

Thanks,
Bobby


[1] 11/3/17 12:03:56 PM org.jgroups.protocols.TP passBatchUp FINER:
localhost-63787: received message batch of 1 messages from localhost-63787
11/3/17 12:03:56 PM org.jgroups.protocols.pbcast.NAKACK2 handleMessages
FINER: localhost-63787: received localhost-63787#1-1 (1 messages)

[2] 11/3/17 11:51:00 AM org.jgroups.protocols.pbcast.STATE_TRANSFER
handleStateReq FINE: localhost-50939: received state request from
localhost-28424