Hello All,



I am a telecommunications engineer who has run into a network trouble when using a communications system simulator on CentOS.

This is how my simulator and network are setup -


- The simulator runs on CentOS 5.5 on VMware.

- I've setup 2 ethernet interfaces ->

eth0 (public) with IP address 148.147.134.XXX and gateway 148.147.134.1. This is connected to my corporate LAN.
eth 1 (private) with IP address 10.30.30.2 and gateway 148.147.134.1.

eth1 has 8000 aliases - eth1:1 to eth1:8000 with IP addresses 10.30.31.1 to 10.30.62.64 respectively.

Each of the eth1 aliases will emulate a virtual phone and are required to register/communicate with a Communications Server which is 148.147.143.XXX. This communication will be over a corporate LAN.

- NOTE: The corporate LAN (148.147.XXX.XXX) does not have a route to 10.30.XXX.XXX and this cannot be done owing to network policies and security.

- NOTE: No configuration changes are possible on the Communications Server (148.147.143.xxx).



My current routing table -

[root@XXXXXXXXX ~]# route
Kernel IP routing table
Destination Gateway Genmask Flags Metric Ref Use Iface
10.30.30.0 * 255.255.255.0 U 0 0 0 eth1
148.147.134.0 * 255.255.255.0 U 0 0 0 eth0
169.254.0.0 * 255.255.0.0 U 0 0 0 eth1
10.16.0.0 * 255.240.0.0 U 0 0 0 eth1
default 148.147.134.1 0.0.0.0 UG 0 0 0 eth0



Problem:

The virtual phones on the eth1 aliases (10.30.31.XXX - 10.30.62.XXX) are unable to register with the Communications Server (148.147.143.XXX). A successful registration will require 4 protocol messages to take place between the phone and the server. This is not happening at the moment.

Is it possible to get this working by using iptables/NAT or any other configuration on the CentOS server?


What I tried so far:

- Enabled IP forwarding.

- Tried some iptables rules, confusing and didnt work, so flushed them all.


Please help!


Thank you

Suresh R

Basics first; are eth0 and eth1 physical devices or is the latter virtual only (everything physically goes through eth0)?

Btw, according to this the last Iface entries should be numbers only, eg '0' instead of 'eth0' and '1' instead of 'eth1'. Can you see if this works?

Thank you Sigg3.net.

eth0 and eth1 are two separate NICs, but they are not really "physical" because the CentOS server runs on VMware. One way I can try and get this to work is to ask my lab admin to create a new switch (virtual) on VMware and link the eth1 NIC to that switch. But doing just that will not suffice. I will have to do a whole lot of configuration and provisioning to route 10.30.31.XXX to the 148.147.XXX.XXX network. That is a very bleak possibility at this time.

On your comment on the eth entries being numbers only - I do not think that is relevant here. I've have systems where eth0 and eth0:1 work fine.