[Users] VE using wrong IP for outgoing connections

[Jan Tomasek]

Hello,

I just realized that my VEs are is using IP address of HW node for
connections initiated from inside of VE, except of their own. For example:

On physical non-OpenVZ system (195.113.1xx.14) I'm pinging my VE
(195.113.2xx.253):

195.113.1xx.14:~# tcpdump -n icmp &
ldap3:~# ping -c 1 195.113.2xx.253
PING 195.113.2xx.253 (195.113.2xx.253) 56(84) bytes of data.
IP 195.113.1xx.14 > 195.113.2xx.253: icmp 64: echo request seq 1
IP 195.113.2xx.253 > 195.113.1xx.14: icmp 64: echo reply seq 1

tcp still running, now I'm pinging from VE (195.113.2xx.253) to physical
system (195.113.1xx.14):

IP 195.113.2xx.254 > 195.113.1xx.14: icmp 64: echo request seq 1
IP 195.113.1xx.14 > 195.113.2xx.254: icmp 64: echo reply seq 1

But instead of .253 I see there .254 - that is IP of HW node. It is same
with TCP and UDP connections. :(

Is there way how to fix this? I'm googling for a hour with no result,
maybe I'm asking wrong questions, networking isn't my strong area :)
Please can someone direct me to right docs? I'm sure this must be asked
many times in past...

Thanks for any help
-- 
-----------------------
Jan Tomasek aka Semik
http://www.tomasek.cz/

-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 189 bytes
Desc: OpenPGP digital signature
Url : http://openvz.org/pipermail/users/attachments/20061204/83f3b890/signature.bin