Network problems

[Matthew Dillon]

:Matthew Dillon wrote:
:
:>     I need a tcpdump from *someone* that monitors the connection that is
:>     stalling.   Let it run a good 2 minutes past the point where the
:>     download stalls.
:> 
:>     tcpdump -n -i <interface> -l host <fetch_target_host>
:
:Matt,
:
:for my hang problems, please look at the two gzip'd tcpdumps in 
:~swildner on leaf (done with only tcpdump -i <interface>). This was 
:without Jeffrey's last patch, though. The hang occurred at around 15:25 
:(but dumping went on after that).
:
:Hope that helps, I have now upgraded to the latest source and am waiting 
:for another hang. :)
:
:Sascha
:
:-- 
:http://yoyodyne.ath.cx

    Ok. It seems to be sending packets out to the internet just fine, but 
    incoming packet traffic stops on the rl0 interface.  At the same time,
    traffic on the internal sis interface seems fine.

    I'm presuming that pppoe is being run by the Tun interface.  It kinda
    looks like everything is going dandy until, poof, data just stops
    coming in over the rl interface.

    It looks like the pppoe link is dying and that ifconfiging tun0 down
    and up again and restarting ppp is reforming the link.   I don't think
    this has anything to do with the MTU issue.  It kinda looks like either
    your DSL box is crapping out or the PPP protocol is allowing the link
    to timeout and causing the DSL box to stop sending packets over it.

    What I would do now is run your PPP demon in debug mode to see what it
    thinks is going on when the link stops working.  I don't think there is
    anything wrong with the rl0 interface.  What I don't understand is why
    the PPP demon isn't restarting the link itself.  It should timeout and
    restart the link automatically.

    It would also be worthwhile to tcpdump the tun0 interface through the
    failure.
     
					-Matt
					Matthew Dillon 
					<dillon at xxxxxxxxxxxxx>