configuring NIC on eth0 - problems

acid_kewpie · 10-22-2001, 12:38 PM

Well can't you even ping the gateway? if you can't then that's just stupid. if you can then i'd guess it's the gateways problem

cliffyman · 10-22-2001, 12:47 PM

I have another linux box, 5 PCs and 11 Macs that can all ping the gateway (which is our router).

Can you think of anything else that might be wrong? Thanks for your help with this, I really appreciate it. Setting up and configuring a network card should NOT be this difficult.

I'm a big fan of Linux but this is extremely disappointing.

Steave · 10-22-2001, 01:25 PM

Quote:

NETMASK=255.255.0.0
NETWORK=192.168.28.0

There's something wrong on this one: if you're using a class B subnet, which means the last two bytes of your ip-address are relevant (192.168.x.x are the addresses you use) Your network should be 192.168.0.0

If you don't, do set your netmask to 255.255.255.0!

Having just some 20 hosts in your network, you should go with a netmask of 255.255.255.0, that is using addresses in the range of 192.168.28.x

Also: what does your routing-table look like now? to ping other hosts on your network you need the following entry:

192.168.28.0 0.0.0.0 255.255.255.0 U 0 0 0 eth0

especially the eth0, that is the device to use is vitally important!

Steave.

cliffyman · 10-22-2001, 01:59 PM

okay,

changed the submask. here's the output of route -n:

Code:

route -n

Kernel IP routing table
Destination     Gateway         Genmask         Flags Metric Ref    Use Iface
192.168.28.85   0.0.0.0         255.255.255.255 UH    0      0        0 eth0
192.168.28.0    0.0.0.0         255.255.255.0   U     0      0        0 eth0
127.0.0.0       0.0.0.0         255.0.0.0       U     0      0        0 lo
0.0.0.0         192.168.28.251  0.0.0.0         UG    0      0        0 eth0

Steave · 10-22-2001, 02:20 PM

Looks pretty much like my working routing-table. Although I don't have the first entry. - I posted this before. But OK, it should work like this, too.

Is your interface up? What kind of an error do you get if you try to ping the gateway?
You should at least not get "network not reachable" with this kind of config. If you do, check the configuration of your NIC. Is it up?

I assume the other boxes in your LAN ping fine. Can you ping the linux-box on 192.168.28.85 from other clients on your LAN?

Steave.

cliffyman · 10-22-2001, 02:27 PM

the interface comes up just fine when i boot the machine. when i ping the gateway i get my standard "Destination Host Unreachable" error.

Unfortunately, none of the other boxes can ping this problem box.

Steave · 10-22-2001, 02:50 PM

OK, we'll probably have to look at another side of the problem. Now that you know everything about routing, I guess the problem is elsewhere.

I just looked at the ifconfig dump you gave in the first place and noticed the 12 errors in TX (Trancieved -> sent) part. Do these kind of errors accumulate when you try to ping other hosts? Do you also get RX errors when you try to ping the box from the outside?

I'd also start checking if the NIC itself is OK. Try to get it to run in some windows-box! If it won't, there's probably some kind of hardware issue.

Are you sure, you're using the right module? ... I think the problem is with the setup of the NIC, not with the network /ip configuration.

Steave.

cliffyman · 10-22-2001, 03:03 PM

hey,

When I ping the outside world and get an error, it does add to the TX errors. But the RX errors never accumulate when I try to ping from one of my other boxes.

I'm pretty certain I'm using the right module. It's a Linksys card and I'm using the newest version of tulip I got from off the web.

The card did function okay when I had Win2000 installed.

Maybe this card simply isn't compatible. I'll have to examine that fact seeing as how it looks like we've exhausted every option here. I really appreciate your assistance, I'll let you know if I find anything else. My patience is wearing very thin so I think I might either buy a known compatible card or try RH 7.1.

Unless you have any other ideas...

Steave · 10-22-2001, 03:29 PM

I don't think its a compatibility problem. It's not RH that runs the card, but the module. And if this is from Tulip and the latest one, there shouldn't be a problem. Unless the guys at Tulip don't know what they're doing

Ok, here is some more to chew on:

the TX-line says carrier=24. Mine says carrier=0. Whatever this means.
I'd think that the carrier values should be the same on all ethernet-cards.

Have you stumbled into any other config-files regarding your network card yet?

If you don't feel like coping with this card under linux any more, I'd suggest you just swap them with one from your win-boxes. (Assuming there are other cards in those...)

Steave.

Hey, I just browsed the ifconfig manpage and found that you can enable/disable arp for ethernet-devices. (the arp-protocol does the translating from ip-addresses to MAC-Adressees)

try "ifconfig eth0 arp" to enable the arp-protocol.

If this really is the error, though, I would send some angry mail to whoever is responsible for not having the card use the arp-protocol in the first place.

cliffyman · 10-23-2001, 03:09 PM

IT WORKS!

After tinkering and rebooting a few more times, I finally got this thing working. And honestly, I have no idea what I did because the config files are the same as what I posted before.

Only one question left: where's the file where I input my DNS info so I can resolve non-local hosts (i can ping everything on my LAN and the outside world fine)

Thank you, everyone who contributed to this thread!

Steave · 10-23-2001, 04:00 PM

yabadoo!

The file you're looking for is

/etc/resolv.conf

the main entries you need to do are:

nameserver x.x.x.x y.y.y.y z.z.z.z ... (space seperated list of NS)
search mynet (will look for name.mynet if name cannot be resolved)
The last entry is useful if you're working in one domain most of the time. (in this case .mynet)
Otherwise it can be ommited.

P.S. Could you do me a favor? I wonder what your carrier line says in the ifconfig output for eth0. Does it say carrier=0?? I wondered if the carrier value should be 0 or else is some kind of an error return code.....

cliffyman · 10-23-2001, 04:11 PM

hey,

thanks, DNS works without a hitch.

and the carrier is 0. thanks again for all your help!

Steave · 10-23-2001, 04:29 PM

One last bit:

I'd check the network cable you are using to connect the linux box, if you encounter any future trouble. carrier=24 might just mean something like no carrier.

Well, guess this is it! Steave.

tekjunky · 11-01-2001, 08:55 PM

in the etc...../ifcfg-eth0 file you have netmask as 255.255.0.0 you might try 255.255.255.0 instead. This is a Class C netmask and works well with the 192.xxx.xxx.xxx range. If you check the others they should be set to this too.

tekjunky

Linux! Never leave /home with-out it!

snooke · 11-15-2001, 11:46 AM

Try using tcpdump to test for network traffic. Like `tcpdump -li eth0`. If you actually have a network connection you'll see all of the traffic on the network. If you see no traffic you may have the wrong module loaded. There are a bunch of variants of the tulip driver. We use DE500's and have to use three different modules depending on the chipset on each NIC.

If tcpdump does show network traffic, start a ping on one tty and monitor tcpdump on another. The evidence about what is happening will be in there. How to interpret it is another matter.....