Re: [circle] Hosts do not find each other

[Nathan Hurst]

On Sun, 14 Sep 2003 17:37:53 -0400 (EDT)
Asheesh Laroia <address@hidden> wrote:

> I am running The Circle on a private network.  These machines cannot
> connect to the primary Circle network.
> 
> (I have tried these procedures on both Linux (Debian) and Windows (XP)
> machines, and with both the latest release (0.35) and today's CVS.)
> 
> So, I started The Circle on one computer, and it waited a minute or
> so, and I started it on a second workstation.  I inputted the IP into
> the manual connect dialog, but neither computer's Circle program
> noticed.  I did "Browse current peers" and it apologizes for not
> finding anyone.
> 
> I've tried leaving the hosts up for a while, but they still don't find
> each other.  I did also try adding the IP of MachineOne to the
> settings.py list of hosts to try at first; MachineTwo still can't find
> MachineOne     on startup, nor on manually adding it through the GUI.
> 
> I do have a lot of Python experience and a little network experience
> (all ad-hoc stuff I've picked up myself, granted).  I noticed the use
> of 127.0.0.1 as a fallback in node.py, and I'm a little worried that
> causes the Circle to bind only to the local loopback address.  Then
> again, I'm sure there's some reasonable explanation for it.
> 
> So, I can't get the Circle to work in this private network.  I would
> *very* much like to use it, and I'm sure the hundreds of people here
> could provide interesting data for your debugging purposes. :-)

That is indeed strange.  The problem we have in testing is getting
access to different environments - circles really well in our full
connectivity, debian environment, but as we move away from that sweet
spot problems appear.

Do you get any hosts at all in your radar screen (top right)?

Does your network do any filtering on UDP (circle only uses UDP for
communications)?

Have you tried connecting the two machines together with a cross-over
cable to make sure there is no filtering occuring at the router?

Which version of python are you running?  there are a few minor bugs to
do with python2.3's new bool type which haven't been fixed (these
usually result in duplicate messages, though, not missed ones)?

Thanks,
-- 
njh
http://www.csse.monash.edu.au/~njh/