[zeromq-dev] ESTABLISHED connection mismatch

Michel Pelletier pelletier.michel at gmail.com
Wed Apr 10 00:11:01 CEST 2013


Are you using their VPC technology or talking over elastic IPs?  The
elastic IP stuff does seem to involve some shenanigans, in that it must
forward or somehow bind the public ip to the internal one.  If you use VPC,
I don't think there's that translation layer involved.

Other than that, I'm out of ideas. :)

-Michel


On Tue, Apr 9, 2013 at 2:51 PM, Garrett Smith <g at rre.tt> wrote:

> No NATing that I'm aware of. But this is EC2, so there could be
> shenanigans afoot. (Shenanigans btw, is the most advanced descriptor
> in my network layer vocabulary.)
>
> The 6 "established" connections on the router-server are still shown
> via netstat -- even after these many hours have passed. If there's a
> timeout in play, it's quite long.
>
> On Tue, Apr 9, 2013 at 3:54 PM, Michel Pelletier
> <pelletier.michel at gmail.com> wrote:
> > Is there NATing between these two hosts? RFC 5382 has some guidelines on
> > idle timeout for established connections, you might be hitting that
> limit.
> >
> > -Michel
> >
> >
> > On Tue, Apr 9, 2013 at 9:36 AM, Garrett Smith <g at rre.tt> wrote:
> >>
> >> I don't know if this is a 0MQ question or a general networking
> >> question. I know that I'm confused.
> >>
> >> I'm troubleshooting some message delivery problems and I've run across
> >> this scenario:
> >>
> >> The ESTABLISHED tcp connections on one server don't match the
> >> corresponding connections on the other server. I'll call one server
> >> "router-server" and the other "dealer-server" (corresponding to the
> >> role and 0MQ socket type one each server).
> >>
> >> On the "router-server" I see these connections to the "dealer-server":
> >>
> >> tcp        0      0 192.168.1.10:1234     192.168.1.11:52726
> >> ESTABLISHED
> >> tcp        0      0 192.168.1.10:1234     192.168.1.11:42556
> >> ESTABLISHED
> >> tcp        0      0 192.168.1.10:1234     192.168.1.11:52728
> >> ESTABLISHED
> >> tcp        0      0 192.168.1.10:1234     192.168.1.11:42774
> >> ESTABLISHED
> >> tcp        0      0 192.168.1.10:1234     192.168.1.11:52727
> >> ESTABLISHED
> >> tcp        0      0 192.168.1.10:1234     192.168.1.11:42563
> >> ESTABLISHED
> >>
> >> On the "dealer-server" I see these to the "router-server":
> >>
> >> tcp        0      0 192.168.1.11:52728     192.168.1.10:1234
> >> ESTABLISHED
> >> tcp        0      0 192.168.1.11:42563     192.168.1.10:1234
> >> ESTABLISHED
> >> tcp        0      0 192.168.1.11:42556     192.168.1.10:1234
> >> ESTABLISHED
> >>
> >> These were listed by "netstat -na" on the respective servers at the same
> >> time.
> >>
> >> This scenario seems to only apply to the problem connection -- the
> >> non-problem router/dealer server pairs have the expected one-to-one
> >> established connections.
> >>
> >> Garrett
> >> _______________________________________________
> >> 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
> >
> _______________________________________________
> 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/20130409/a4ba0707/attachment.htm>


More information about the zeromq-dev mailing list