Yep, I realised after I sent it,
my packets were dropped when I had something I thought was similar.

Sorry for the noise.

On Fri, 2004-12-17 at 01:01 +0000, Nix wrote:
> On Thu, 16 Dec 2004, Antoine Martin spake:
> > Looks to me like this is a routing problem, check your arp tables and
> > your routes. It may be that something is occasionally causing a packet
> > to come out of the wrong interface/wrong mac, which confuses the lower
> > network layers.
> 
> Alas, I don't think it's that simple :(
> 
> My arp tables are simple; virtually everything has the MAC of the ADSL
> router.
> 
> Every packet going out of the wrong interface gets syslogged, and I see
> no such syslog messages, just heaps of NFS timeout messages.
> 
> (See <http://www.esperi.demon.co.uk/nix/iptables> for my firewall rules;
> a few sensitive external IP addresses have been elided. Look at the log
> rules therein.)
> 
> ip addr:
> 
> 1: lo: <LOOPBACK,UP> mtu 16436 qdisc noqueue 
>     link/loopback 00:00:00:00:00:00 brd 00:00:00:00:00:00
>     inet 127.0.0.1/8 brd 127.255.255.255 scope host lo
> 2: teql0: <NOARP> mtu 1500 qdisc noop qlen 100
>     link/void 
> 3: dummy0: <BROADCAST,NOARP> mtu 1500 qdisc noop 
>     link/ether 00:00:00:00:00:00 brd ff:ff:ff:ff:ff:ff
> 4: gordianet: <BROADCAST,MULTICAST,ALLMULTI,UP> mtu 1458 qdisc pfifo_fast 
> qlen 100
>     link/ether 00:60:97:79:e2:c1 brd ff:ff:ff:ff:ff:ff
>     inet 192.168.14.1/24 brd 192.168.14.255 scope global gordianet
> 5: adsl: <BROADCAST,MULTICAST,ALLMULTI,UP> mtu 1458 qdisc cbq qlen 10
>     link/ether 00:60:97:79:e2:c1 brd ff:ff:ff:ff:ff:ff
>     inet 194.247.41.52/24 brd 255.255.255.255 scope global adsl
>     inet 192.168.14.159/32 scope global adsl
> 
> 
> ip route:
> 
> 192.168.14.160 dev adsl  proto static  scope link  src 192.168.14.159 
> 194.247.41.0/24 dev adsl  proto kernel  scope link  src 194.247.41.52 
> 192.168.14.0/24 dev gordianet  proto kernel  scope link  src 192.168.14.1 
> 224.0.0.0/4 dev gordianet  scope link 
> default via 194.247.41.52 dev adsl 
> 
> 
> (note an arcaneness: the ADSL router is on 192.168.14.160, the same
> subnet as all the stuff on the other interface: it's accessed via a
> different source address (to avoid being masqueraded).
> 
> 
> ... hm. Some more info. This is definitely an interface-specific thing.
> The blockages appear to affect different interfaces at different times:
> i.e., one interface freezes, in both directions --- and it's not just
> one destination, it's everyone except for myself. e.g., pinging the
> broadcast address on the local net:
> 
> 64 bytes from 192.168.14.1: icmp_seq=343 ttl=64 time=9.7 ms
> 64 bytes from 192.168.14.16: icmp_seq=342 ttl=64 time=5010.4 ms (DUP!)
> 64 bytes from 192.168.14.14: icmp_seq=342 ttl=64 time=5011.0 ms (DUP!)
> 64 bytes from 192.168.14.18: icmp_seq=342 ttl=64 time=5011.5 ms (DUP!)
> 
> I'm starting to suspect the host's TUN/TAP here, y'know: the fact that
> it's a simultaneous bidirectional freeze, and that the packet to myself
> (automatically sent directly to myself by the kernel) is *not* delayed
> is possibly significant.
> 
> (But there were no TUN/TAP changes between 2.4.26 and 2.4.28. Damn.
> Maybe it's TCP Vegas or something? *clutches at straws in
> ChangeLog-2.4.2[78]*)
> 



-------------------------------------------------------
SF email is sponsored by - The IT Product Guide
Read honest & candid reviews on hundreds of IT Products from real users.
Discover which products truly live up to the hype. Start reading now. 
http://productguide.itmanagersjournal.com/
_______________________________________________
User-mode-linux-user mailing list
[EMAIL PROTECTED]
https://lists.sourceforge.net/lists/listinfo/user-mode-linux-user

Reply via email to