[zeromq-dev] What is the exact issue with enabling MULTICAST_LOOP on ZMQ sockets?

[Parag Patel]

Does multicast loopback work without known issues on linux (regardless of java or not)?

From: zeromq-dev-bounces at lists.zeromq.org [mailto:zeromq-dev-bounces at lists.zeromq.org] On Behalf Of Steven McCoy
Sent: Saturday, June 08, 2013 9:03 AM
To: ZeroMQ development list
Subject: Re: [zeromq-dev] What is the exact issue with enabling MULTICAST_LOOP on ZMQ sockets?

On 8 June 2013 08:00, Pieter Hintjens <ph at imatix.com<mailto:ph at imatix.com>> wrote:
If anyone really wants the MULTICAST_LOOP functionality, we can add it
back and try to make the documentation and defaults clear...


I can make things even more complicated that if both publisher and subscriber are Java applications then it will never work.

The problem stems from PGM requiring unicast capability for NAKs and how packets are routed when you are listening on the same port.  For PGM on Java on Windows a unicast NAK be will routed to the receive multicast socket without hitting the IP stack and any sending application on the host.

One might think IBM's method of splitting unicast and multicast encapsulated ports may help: the loop only applies to one socket.  Sockets are used for sending and receiving such that multiple binding conflicts thus occur.

The preferred resolution is to add an IPC/TCP bind to your PGM 0mq socket, i.e.

s = zmq_socket (context, ZMQ_SUB);
zmq_bind (s, "pgm://239.192.0.1<http://239.192.0.1>;7500");
zmq_bind (s, "ipc:///tmp/loop/7500");
...

--
Steve-o
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.zeromq.org/pipermail/zeromq-dev/attachments/20130610/ca50f400/attachment.htm>