From: Dj Statik <djstatik@dj...> - 2003-10-31 22:20:57
Am running in to difficulties trying to get UML networking to work.
The host is running a 2.4.22 kernel with the SKAS patch applied. UML is
running 2.4.22 with the UML patch applied. Both systems have tun/tap
enabled in the kernel.
My network is set up as follows. The host system has an ip for example of
22.214.171.124 and my UML has an ip of 126.96.36.199
188.8.131.52 is assigned to eth0 on the host also
Here is the script I am using to bring up the UML.
/usr/sbin/tunctl -u djstatik
/sbin/ifconfig tap0 184.108.40.206 netmask 255.255.255.255 up
/sbin/route add -host 220.127.116.11 tap0
/bin/echo 1 > /proc/sys/net/ipv4/conf/tap0/proxy_arp
/usr/sbin/arp -Ds 18.104.22.168 eth0 pub
su djstatik - -c "/usr/local/uml/linux-2.4.22-uml defs=mount
User djstatik is in a group called uml which has write access to
When the host comes up I do a
ifconfig eth0 22.214.171.124 up
and I am able to:
* Ping the host from the UML
* ping the outside world from the UML
* ping the UML from the host.
but I can't ping the UML from the outside world.
I suspect this to be a routing problem on the host, but I honestly can't
spot what the problem is.
If anyone can see anything painfully obvious that I am doing wrong then I
would really appreciate your suggestions.
> Am running in to difficulties trying to get UML networking to work.
I know this is my standard reply about this, but please try my howto:
It avoids the need for arp/routing on the host or UML.
> and I am able to:
> * Ping the host from the UML
> * ping the outside world from the UML
> * ping the UML from the host.
> but I can't ping the UML from the outside world.
Do the pings from outside->uml time out, or do they receive some kind of
I don't think it is a routing problem if you can ping uml->outside but not
outside->uml because the replies are being routed successfully.
Are you sure it isn't a firewall issue? Try 'iptables -t filter -F' on all
Is NAT involved?
Are you sure the ping command isn't sitting around waiting for DNS lookups
and only appearing to timeout? Try 'ping -n wherever'.
I would suggest using the tcpdump command to try and debug this. Do 'tcpdump
-i ethX/tunX -n icmp' on each involved machine while pinging. Try to spot
where the pings are being lost.