[Time] Name | Message |
[00:37] ADruz
|
mikko__: thanks
|
[01:12] mike8901
|
How should I go about determining the proper value of io_threads?
|
[01:47] mike8901
|
Is there any way to have multiple clients connect to a single TCP port(but be represented by multiple unicast sockets) with zmq?
|
[04:31] rmancy
|
Is this the best place to ask for help with my zeromq app?
|
[04:31] mike8901
|
I guess so.
|
[04:35] rmancy
|
ok great :-) I'm trying to make a simple server/client with ZMQ_REP/ZMQ_REQ respectively. The server does the recv(), then send() seemingly fine, however my client after doing it's initial send() to server (which the server gets), it sits there waiting for recv(). Is REP and REQ appropriate here ?
|
[04:36] guido_g
|
yes, what did you expect?
|
[04:37] rmancy
|
guido_g, Well I'm expecting the client to actually recv the data that server sends back...however it doesn't seem to receive it. Just seems to sit there _waiting_ to receive something, but doesn't actually receive it
|
[04:37] rmancy
|
Have I missed something ?
|
[04:37] guido_g
|
w/o seeing the code, hard to tell
|
[04:38] guido_g
|
is there a packet filter in the way?
|
[04:38] guido_g
|
aka firewall
|
[04:38] rmancy
|
guido_g, hmmmm.
|
[04:38] rmancy
|
will check
|
[04:41] rmancy
|
guido_g, nope
|
[04:42] rmancy
|
I'm running zmq_forwarder with a basic 'in' and 'out' elements that obviously work on the initial send, just not on the response it seems
|
[04:42] guido_g
|
did you compare your code against the cookbok example?
|
[04:42] guido_g
|
ouch
|
[04:43] guido_g
|
as far as I understood, req/rep is meant for direct connection between the requesting client and the services
|
[04:43] rmancy
|
guido_g, oh ok. So no forwarder? I had used the forwarder from a SUB/PUB example, and just assumed I still needed it. Perhaps not though !
|
[04:44] guido_g
|
not only perhaps
|
[04:44] guido_g
|
see http://www.zeromq.org/docs:cookbook
|
[04:44] guido_g
|
a forwarder spreads out the messages to all subscribed entities
|
[04:44] guido_g
|
req/rep works differently
|
[04:51] rmancy
|
guido_g, thanks. Didn't realise :-(
|
[04:51] rmancy
|
I'm also not using 'localhost' and 'lo' as I get errors creating sockets on those. Using my public IP instead, but I don't think that should have a bearing
|
[04:52] rmancy
|
or possible it does
|
[04:52] guido_g
|
try 127.0.0.1, might be a naming issue
|
[04:57] guido_g
|
i'm off, laters
|
[04:58] rmancy
|
guido_g, thanks, with that help i should be able to nut it out :-)
|
[06:58] guido_g
|
re
|
[12:20] rmancy
|
Hi, anyone able to help with a simple zeromq prob I have ?
|
[12:55] sustrik
|
rmancy: hi
|
[12:55] sustrik
|
what's the problem?
|
[14:59] ka3ak
|
hello
|
[15:00] ka3ak
|
can someone help me out with a zmq issue
|
[15:00] sustrik
|
ka3ak: hello
|
[15:00] ka3ak
|
hi
|
[15:01] sustrik
|
what's the problem?
|
[15:01] ka3ak
|
i have a simple server set up with a ZMQ_REP socket ...
|
[15:01] ka3ak
|
once in a while the recv will hang
|
[15:02] ka3ak
|
the client send an enourmous amount of messages, so i'm not sure if the queue is getting backed up or what
|
[15:02] sustrik
|
nope
|
[15:02] sustrik
|
looks like a bug
|
[15:02] ka3ak
|
i've tried increasing the high water mark and swap, but nothing works
|
[15:03] sustrik
|
can you send the test program to the mailing list?
|
[15:03] sustrik
|
or open an issue in a bug tracker?
|
[15:03] ka3ak
|
sure
|
[15:03] sustrik
|
thx
|
[15:03] ka3ak
|
just to give you a bit more info on the hack to get around this ...
|
[15:04] ka3ak
|
while (1) { message_t req; if (!sock_.recv(&req, ZMQ_NONBLOCK)) continue; } but this will spike up the cpu to 100 for a second or so for obvious reasons
|
[15:05] ka3ak
|
but this fixes the problem albeit with a short pause and cpu spike on the server side of things
|
[15:05] sustrik
|
ok, i see
|
[15:36] CIA-15
|
zeromq2: 03Martin Sustrik 07master * rad6fa9d 10/ (src/rep.cpp src/req.cpp src/xrep.cpp src/xrep.hpp): initial version of multi-hop REQ/REP - http://bit.ly/cbReUI
|
[19:38] mike8901
|
If I send a message, and the message gets deallocated immediately after I send the message, will the message still go through OK?
|
[19:38] mike8901
|
i.e. will send copy the message?
|
[19:45] sustrik
|
mike8901: it won't copy the message, it will take possession of it
|
[19:46] sustrik
|
there's nothing left for you to deallocate
|
[19:53] mike8901
|
so I can allocate a message_t on the stack, copy the data into the message, and send it, without worrying about what happens to the data?
|
[20:00] sustrik
|
yes
|
[20:04] mike8901
|
ok, cool :)
|
[21:40] bgranger
|
sustrik: I am going to be working on the Python bindings some more and have a few questions about release schedules...
|
[22:20] CIA-15
|
pyzmq: 03Brian E. Granger 07master * r506679b 10/ (6 files in 4 dirs): Renaming P2P -> PAIR. - http://bit.ly/9MC0TY
|
[22:22] CIA-15
|
pyzmq: 03Brian E. Granger 07master * r5177246 10/ (5 files in 2 dirs): Finish renaming of P2P -> PAIR in filenames. - http://bit.ly/92CZZQ
|
[22:36] CIA-15
|
pyzmq: 03Brian E. Granger 07master * rbb7c339 10/ (zmq/_zmq.c zmq/_zmq.pyx zmq/tests/test_poll.py): Removed stopwatch stuff and fixed select. - http://bit.ly/9GT4M8
|
[22:39] CIA-15
|
pyzmq: 03Brian E. Granger 07master * r16b0c20 10/ setup.py : Adding zmq.tests to setup.py packages. - http://bit.ly/ceobEw
|
[23:26] mikko
|
sustrik: hi
|
[23:26] mikko
|
http://pastie.org/938296
|
[23:26] mikko
|
ending up to this with git HEAD
|
[23:31] mikko
|
ah
|
[23:31] mikko
|
my mistake
|