[zeromq-dev] Fwd: zeromq 4.0.4 zeromq crash in async router dealer message pattern

John Thomas jtbeta at gmail.com
Sun Aug 2 16:48:17 CEST 2015


We have implemented a service that utilizes the async server. I am using
zmq_proxy with a ROUTER front end and DEALER backend. I have a bunch of
workers connected to this DEALER backend via DEALER

Sockets. The core loop in each worker receives messages via DEALER socket
and performs a service based on the message and sends a reply back.

Clients connects to the ROUTER front end at well know ip:port address bound
by the ROUTER sockets and sends messages to perform some action by the
server. Servers dispatch the requests to the workers via the zmq_proxy
thread. The workers process the message and respond back to client. The
clients and servers are running in different boxes and they communicate
over TCP.

We are being hit by the following crash in the client side while the client
is waiting to receive a response for a message from server.  The client is
also opening a DEALER socket to server (ROUTER) socket for receiving
replies from server.

The stack trace of the crash is given below.

#13 0x00007ffff7563b59 in zmq::zmq_abort (errmsg_=errmsg_ at entry=0x7ffff758bb85
"errno == EAGAIN") at err.cpp:74

#14 0x00007ffff75761f5 in zmq::socket_base_t::process_commands
(this=this at entry=0x7ff975bdc150, timeout_=timeout_ at entry=0,
throttle_=throttle_ at entry=false) at socket_base.cpp:911

#15 0x00007ffff757675c in zmq::socket_base_t::recv (this=0x7ff975bdc150,
msg_=msg_ at entry=0x7ffff6242bd0, flags_=<optimized out>) at

#16 0x00007ffff7589949 in s_recvmsg (s_=<optimized out>,
msg_=0x7ffff6242bd0, flags_=<optimized out>) at zmq.cpp:460

This is occurred with zeromq-4.0.4.

What could be the reason for this crash? Any thoughts?

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.zeromq.org/pipermail/zeromq-dev/attachments/20150802/81d990d9/attachment.htm>

More information about the zeromq-dev mailing list