You don't assign gateways on a per-NIC basis, you have one routing table for the entire system, which contains a number of gateways, one of which is the default gateway. What you're thinking of is ip-forwarding, but this is not set in the routing table. What does the faulty routing table look like after a reboot? regards Anders On Tuesday 06 November 2001 06.15, W.D.McKinney wrote:

Hi Kevin,

My modem is a cDSL IMeg Nortel unit, ISP provided, and it connects to a Redback, it's a PPPoE connection, so I'm using roaring penguin which is included in the SuSE disro. I have a single static routeable IP address, and the rest are based on the 192.168.xxx.xxx address scheme. The modem is not anything I can access as the connection is based on PPPoE. The "whole" problem is software based not hardware.

As the former IP Operations Manager of the ISP I connect to I quite familiar with Cisco, etc., and routing. The issue is that if I set the parameters for networking in YaST, they do not update the system properly.

If I give eth0 an address of 209.193.48.40 and no gateway, it should not have the same address in the gateway field, period. But that's exactly what happens when I add eth1 with 192.168.0.1 for it's address and 209.193.48.40 as it's gateway. It appends the gateway address from eth1 to eth0 also. Is it impossible for YaST to have 2 fifferent gateways ?

Comprende ?

/Dee

Kevin L Hochhalter wrote:

...

On Monday 05 November 2001 18:54, W.D.McKinney wrote:

...

Please help.

My gateway/www/firewall running 7.3 Pro has 2 NIC's :

00:0f.0 Ethernet controller: 3Com Corporation 3c905B 100BaseTX [Cyclone] (rev 30) 00:11.0 Ethernet controller: 3Com Corporation 3c905B 100BaseTX [Cyclone] (rev 30)

Both work fine right now.

The network is based on a static Ip and looks like this:

ISP ----->DSL Modem--->NIC #0 (209.193.48.40)

Internal network is :

Gateway Firewall web server ___________|_________________________

(switch) MP3 server workstation

(switch#2)

|__________________________________________________________________ |_ | | | | | | | | | etc.

PC PC PC PC PC PC PC PC

Now when I set the network on the gateway/firewall/everything server, I give eth0 a static IP and DNS, but no gateway as this what my ISP sez due this being a DSL connection. For the internal network, I give eth1 a 192.168.xxx.xxx address, a gateway of 209.193.48.40, mask of 255.255.255.0 and this works fine.

Looks like this when it works:

Kernel IP routing table Destination     Gateway         Genmask         Flags Metric Ref    Use Iface 209.193.48.254  0.0.0.0         255.255.255.255 UH    0      0        0 ppp0 192.168.0.0     0.0.0.0         255.255.255.0   U     0      0        0 eth1 209.193.48.0    0.0.0.0         255.255.255.0   U     0      0        0 eth0 0.0.0.0         209.193.48.254  0.0.0.0         UG    0      0        0 ppp0

Except when I need to reboot for whatever, the network re-configures itself. It gives the gateway address of eth1 to eth0 as well, which does not work obviously. Stopping both the internal IP connection and the external IP connection. So what's the fix ? Having to be in front of this server to fix the networking after a power outage, etc., is not feasable.

By the way, 6.2 never produced this problem. I'd hate to bump back to an older version as this one is running 7.3 Pro.

Any ideas will be appreciated.

Regards

Well, interesting problem. My first question is, what brand of dsl modem are you using? My second question is, when you leased the block of addresses, how many did you get? You will often get a block of eight or so, and one of those is almost always the address that the modem will take whenever it reboots. I imagine this depends on the modem brand, though, which is why I asked. My set up is very similar to yours, and I just set the default gateway of both nic's to the address of the modem. The address of the modem has nothing to do with the addresses that you assign to your nic's, other than acting as the gateway. Nobody should even be able to ping your modem. Get access to your modem's operating system, and see what address it is set to. It should be one of the addresses in the block that you leased. You can usually connect to your modem using ckermit if you have a serial controller cable for it. If you don't, try getting in with telnet on each of the addresses that your isp leased you. Once you have that address, set the gateway of both nic's to that, and it ought to work. Well, at least it works with my isp :-)

If you happen to have a Cisco modem (I have a Cisco 678), let me know, and I can hopefully lend a hand with getting access to the operating system.

Good Luck, Kevin

On Tuesday 06 November 2001 06.34, W.D.McKinney wrote:

I have a NIC connected to a DSL model with a provided staic IP.

I have a 2nd NIC in the same box, that uses the static address of the first NIC as it's gateway.

I don't know what this means. A NIC doesn't have a gateway. A routing table has. In your first mail you sent a working routing table

Destination Gateway Genmask Flags Metric Ref Use Iface 209.193.48.254 0.0.0.0 255.255.255.255 UH 0 0 0 ppp0 192.168.0.0 0.0.0.0 255.255.255.0 U 0 0 0 eth1 209.193.48.0 0.0.0.0 255.255.255.0 U 0 0 0 eth0 0.0.0.0 209.193.48.254 0.0.0.0 UG 0 0 0 ppp0

The gateway field means that if there is a valid adress there, the packets are to be sent there for further routing. If it is, as it is here, 0.0.0.0, they'll just be transmitted to the interface without alteration. If both gateway and iface are empty/invalid, you have a corrupt routing table. If your routing table has (and from your description it sounds very much so) 209.193.48.40 in the gateway field of the entry that has eth1 in the iface column, then this cannot ever work, since there is no such ip number on the other side of that interface. This functionality is what ip forwarding is for. And from the routing table you give that I quoted above, it looks like it is simply being ignored. Now, here all packages that match 209.193.48.0/24 are sent to eth0, all packages that match 192.168.0.0/24 are sent to eth1, and all packages that don't match either are sent to the ppp0 interface. How does the routing table differ after a reboot?

I am not ip-forwarding yet in this configuration.

The routing table in SuSE is /etc/route.conf which is set by YaST. If I input into YaST, it should be translated in /etc/route.conf. This does not happen correctly.

Then what happens? It would be easier to solve if you post the numbers instead of trying to describe it. regards Anders