[zeromq-dev] test_proxy hanging occasionally

[Brian Silverman]

I'm pretty sure I found the problem: the sockets in the test had ZMQ_LINGER
set to the default (-1, aka forever). Setting them all to 0 seems to have
fixed the hanging problem. It ran 100,000 times over the weekend without
hanging. I had come across that potential solution before sending the
email, but realized just after asking for help that I missed setting the
option on some of the sockets...

However, with the ZMQ_LINGER fix I did get an assertion failure in lb.cpp
at the "zmq_assert (!more);" line. I'm not sure if that's related or not.
That happened in 1 out of the 100,000 runs. Working on chasing that down a
bit more now.

What's the appropriate path forwards here? Should I send a github pull
request for setting ZMQ_LINGER, or does it make more sense to figure out
the assertion failure first, or something else?

Thanks,
Brian

On Sat, Jul 18, 2015 at 2:05 AM, Pieter Hintjens <ph at imatix.com> wrote:

> It could be timing dependent. To track this down the first thing would
> be to run the test in isolation, and see if you can get it to fail.
> Then, to change the timers (I see it waits for 200msec between
> messages) to see if that makes it fail more often.
>
> In any case we need to find out where it's hanging...
>
> -Pieter
>
> On Sat, Jul 18, 2015 at 3:16 AM, Brian Silverman <brian at peloton-tech.com>
> wrote:
> > Hi,
> >
> > I'm seeing test_proxy hanging occasionally (on the order of 1 in 100,000
> > runs). It never returns a non-zero exit code though. I'm doing this on
> > Debian Jessie with the default 3.16.0-4-amd64 kernel. I'm using commit
> > 2e5435e38f91a33ed2f9b78c046a3e7b488a453f. Any ideas?
> >
> > Thanks,
> > Brian Silverman
> >
> > _______________________________________________
> > zeromq-dev mailing list
> > zeromq-dev at lists.zeromq.org
> > http://lists.zeromq.org/mailman/listinfo/zeromq-dev
> >
> _______________________________________________
> 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/20150720/9c9ec955/attachment.htm>