[zeromq-dev] Interrupted system call

MinRK benjaminrk at gmail.com
Thu Jul 28 15:21:50 CEST 2016


On Thu, Jul 28, 2016 at 2:28 PM, Frédéric <fma at gbiloba.org> wrote:

> Le jeudi 28 juillet 2016, MinRK a écrit :
>
> > Following the behavior of Python 3.5, interrupted system calls are
> > retried, starting in pyzmq 14.7.
>
> I tried added a manual retry, but this does not help.
>
> > Are you creating the sockets after forking with multiprocessing? You
> > shouldn't use any sockets that were created in the host process in any
> > of the forks.
>
> Mmm, no, I don't instanciate a new socket :o/
>
> I have to figure out how to do that with my Singleton logger and
> multiprocessing module...
>
> This is strange that it work fine a simple example, or even in my complete
> application on my desktop PC, and fail on Rpi... What can cause such
> behaviour difference?
>
> BTW, could someone resume what is needed to re-create or not in a
> multi-threaded and multi-processing context? Is ok to share Context? When
> do I need to use zmq.Context() or zmq.Context.instance() ?
>

You must create a new `zmq.Context()` in each process.


>
> Thanks,
>
> --
>     Frédéric
> _______________________________________________
> zeromq-dev mailing list
> zeromq-dev at lists.zeromq.org
> http://lists.zeromq.org/mailman/listinfo/zeromq-dev
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.zeromq.org/pipermail/zeromq-dev/attachments/20160728/375744ed/attachment.html>


More information about the zeromq-dev mailing list