Hi Lee,
The PUB socket is neutral towards SUBscribers that's why it's a publisher.
Depending on what you are trying to achieve with your mesh you may need to
use different socket pair if you need such "filtering". Did you try
XPUB/XSUB?
On Sun, Apr 14, 2013 at 10:47 PM, Lee Sylvester wrote:
>
Hi Asif,
I'm hoping Divya will provide us with reusable tests that show the problem.
-Pieter
On Mon, Apr 15, 2013 at 7:50 AM, asif saeed wrote:
> Excellent, Divya. Very good analysis.
>
> @Pieter,
>
> I am also going to use ZeroMQ in an Ubuntu environment and Divya's analysis
> is really wo
On Mon, Apr 15, 2013 at 8:02 AM, crocket wrote:
> If I call Read Tag from DSCTagTest via ZeroMQ, it takes 350~370ms to call
> Read Tag 1000 times.
This is doing a round-trip, right?
That becomes slow. There's a lot of latency for pushing single
messages through the whole stack like this.
It's
Hi,
I saw the XSUB and XPUB; they're dealer / router related, right? I found it
hard to find information on using these directly. If I swap out PUB/SUB for
XPUB/XSUB, how then do I specify who can receive what packets?
Thanks,
Lee
On 15 Apr 2013, at 08:03, "A. Mark" wrote:
>
> Hi Lee,
>
> This is doing a round-trip, right?
In the java program, I have a DEALER socket named "tagReader", and I use it
as below.
for(String addr : addrArray)
tagReader.send(addr.getBytes(charset));
for(int j=0; j___
zeromq-dev mailing list
zeromq-dev@lis
On Mon, Apr 15, 2013 at 9:31 AM, Lee Sylvester wrote:
> I saw the XSUB and XPUB; they're dealer / router related, right? I found it
> hard to find information on using these directly. If I swap out PUB/SUB for
> XPUB/XSUB, how then do I specify who can receive what packets?
XPUB/XSUB are diffe
In other words, I send 1000 requests and then retrieve 1000 results later.
Each message from DEALER to ROUTER causes LabVIEW to call "Read Tag" and
send back the result to DEALER.
Is it still a round-trip?
___
zeromq-dev mailing list
zeromq-dev@lists.ze
On Mon, Apr 15, 2013 at 9:56 AM, crocket wrote:
> In other words, I send 1000 requests and then retrieve 1000 results later.
> Each message from DEALER to ROUTER causes LabVIEW to call "Read Tag" and
> send back the result to DEALER.
> Is it still a round-trip?
No, this would be batched.
I'd tr
Thank you, Pieter. I'll look there.
Regards,
Lee
On 15 Apr 2013, at 08:55, Pieter Hintjens wrote:
> On Mon, Apr 15, 2013 at 9:31 AM, Lee Sylvester
> wrote:
>
>> I saw the XSUB and XPUB; they're dealer / router related, right? I found it
>> hard to find information on using these directly.
Hi Pieter,
So, am I right in saying that, unless I create specific connections to each
node, I can't limit my traffic by node? If so, is there any way to create
specific connections to nodes but using the same port? I like the idea of
using one port for outgoing messages and one for incoming
Hi,
I am running zeromq in pgm mode. I'm seeing large amount of msg loss and
very stuttery msg flow. Is this to be expected or am I doing something
wrong.
FYI I'm running a simple pub sub in pgm mode.
Thanks
Susan
___
zeromq-dev mailing list
zeromq-dev
Lee,
I've updated the FAQ to answer your question since it's becoming quite
a common trend.
http://www.zeromq.org/area:faq#toc0
Basically, you're trying to run before you've learned to wear these
new shoes. It usually ends poorly. It's better to read the Guide until
you really do understand what
Hi Pieter,
I've actually bought and read the book. My issue isn't with understanding how
it works, but with simply filtering on the publisher end. The pattern I need
is very simple and my needs are simple.
Anyhow, I've decided to go back to using RabbitMQ, as its direct messaging
flags are e
On Mon, Apr 15, 2013 at 1:13 PM, Lee Sylvester wrote:
> I've actually bought and read the book. My issue isn't with understanding
> how it works, but with simply filtering on the publisher end. The pattern I
> need is very simple and my needs are simple.
>
> Anyhow, I've decided to go back to
Thank you, Pieter, that's exactly what I wanted to hear :-)
Regards,
Lee
On 15 Apr 2013, at 12:34, Pieter Hintjens wrote:
> On Mon, Apr 15, 2013 at 1:13 PM, Lee Sylvester
> wrote:
>
>> I've actually bought and read the book. My issue isn't with understanding
>> how it works, but with simp
Hi,
I'm working on a some code that generates a large volume of data and I'm
using zmq to push the data off to a forked process for further work. I've
run into a weird situation where the child process which is trying to
receive the messages doesn't receive any until the main process stops
produci
On 15 April 2013 06:05, Susan Tharakan wrote:
> I am running zeromq in pgm mode. I'm seeing large amount of msg loss and
> very stuttery msg flow. Is this to be expected or am I doing something
> wrong.
>
>
The symptoms suggest the reliability of the socket is breaking and thus you
are seeing res
You should check the return value of those calls to zmq_bind, zmq_connect
etc. Try 127.0.0.1 instead of localhost, you're probably getting ENODEV on
the bind.
Ian
On Mon, Apr 15, 2013 at 1:03 PM, Lucas Maxwell wrote:
> Hi,
>
> I'm working on a some code that generates a large volume of data an
Hi,
I need to connect through a SOCKS proxy. What's the best way to do this? I
suspect I'll need to use ZMQ_ROUTER_RAW, talk socksy to it, then switch raw
back off. Only problem is we're using 3.2.2 as it's the latest stable
release, which doesn't include the ZMQ_ROUTER_RAW option.
Any suggestion
Hi Pieter,
Please find attached the test code.
*Test Description: *
send_process: - Gets the time before sending a message and sends it as part
of the message.
- Message send continuously with a delay of 1 ms.
- Uses ZMQ_PUB socket and IPC protocol.
I have an application that currently sends msgs via PUSH and a worker that
gets them via PULL. but i want multiple workers to get the same data so
thinking of PUB/SUB but concerned about msg loss.
Assuming sufficiently high watermark, is there a chance msgs could be
dropped/lost in a pub/sub scena
On Mon, Apr 15, 2013 at 7:25 PM, Divya Mohan wrote:
> Test ran for more than 48hrs.
> NOTE: Message delay will be seen only after running the test for a couple
> of hours. (in my case after 10 hrs)
Thanks for the test code.
Is there a particular reason you're using multiple threads in the
rece
On Mon, Apr 15, 2013 at 7:07 PM, Evan Gates wrote:
> I need to connect through a SOCKS proxy. What's the best way to do this? I
> suspect I'll need to use ZMQ_ROUTER_RAW, talk socksy to it, then switch raw
> back off. Only problem is we're using 3.2.2 as it's the latest stable
> release, which do
> I'd try more messages, see if the time is proportional (10K, 100K,
> 1M). I'd then replace the LabView piece with null code (doing no work)
> so you can see which side the time is being taken.
It turned out that labview-zmq(LabVIEW binding) itself was slow.
labview-zmq is just a wrapper around
JeroMQ is going to be
slower than the JZMQ.
Joshua
crocket
April 15, 2013
8:22 PM
> I'd try more
messages, see if the time is proportional (10K, 100K,> 1M). I'd then
replace the LabView piece with null code (doing no work)
> so you
can see which side the time is being
That is exactly it. I get an ENODEV which appears to be 'operation not
supported by device' for OSX. I thought ENODEV was that the device didn't
exist at all?
Thanks for your help.
On 16 April 2013 01:00, Ian Barber wrote:
> You should check the return value of those calls to zmq_bind, zmq_con
Looking into the zmq_tcp/zmq_bind doco is only accepts the numeric form,
interface name, or wildcard when binding. DNS names are ok when connecting.
Looking through the OSX errno.h is appears that ENODEV is still 19 in line
with POSIX but the text version of the error indicates that you can't do
th
Hi,
sorry for my poor English.
My program is the ROUTER model (connecting 2 DEALER programs via unix
domain socket) on linux x86_64,work with tornado ioloop.
I registered a clean function for the sigterm, and in this function i call
the ioloop.stop() ,stream.close() and context.destroy().
I found t
Hi Pieter,
Please see my answers inline.
Is there a particular reason you're using multiple threads in the
receiver? Also, why are you creating a pub socket there?
- Actual application requires multiple threads for handling different
functions. Socket* will not be* shared between different thread
i have network client / server application that using java zeromq
i have the main server and N clients that polls the server.when the server
gets online.
the clients connect him and there some short massaging going on between them
until now i done with single client and it worked fine.
but when ad
30 matches
Mail list logo