Network adapter down

Louis_Carole · 09-14-2009, 05:12 PM

Dear All,

A node on a cluster that our lab has had sitting in a dusty room since time forgotten does not seem to want to talk to the rest of the world any more.

Following is the information I found was being asked for on other posts for similar questions, transcribed by hand.

Thanks in advance,

Louis

PS Hardware documentation? Hahahahahahahaha!

Code:

# udev -a   # NOTE: No distro info
Linux 2.4.20-8 #1 Thu Mar 13 17:18:24 EST 20033/13/2003 i686 Athlon i386 GNU/Linux

# lspci
...
Realtek Ethernet Controller, yada yada
...

# modprobe e1000
...
init_module: No such device
...

# dmesg | tail   # nothing of interest before, I don't think.
...
Intel PRO/1000 Network Driver version 4.4.19-K1
sis900.c: v1.08.06 9/24/2002
sis900.c: v1.08.06 9/24/2002
sis900.c: v1.08.06 9/24/2002
Intel PRO/1000 Network Driver version 4.4.19-K1

# cat /var/log/boot.log | grep fail
Mounting NFS filesystems: Failed
Bringing up interface eth0: failed
attempting to contact yp server: failed
attempting to contact yp server: failed

# lsmod
...
nfsd
...

# ifconfig eth0 whatever
...
sis900 device eth0 does not seem to be present, delaying initialization
...

# ifup eth0
...
SIOCSIF{ADDR,NETMASK,BRDADDR}: No such device
eth0: unknown interface
...

# grep -R eth0 /etc/*
...
/etc/modules.conf:alias eth0 sis900
/etc/radvd.conf:alias eth0 sis900
/etc/sysconfig/network-scripts/ifcfg-eth0:DEVICE=eth0
...

MS3FGX · 09-14-2009, 05:29 PM

Well the first question, and I am going to assume you already have an answer to this, would be why are you running such an ancient kernel? More than likely the problem is the outdated and possibly buggy drivers that your kernel is using. There is almost no doubt in my mind that using a modern kernel would resolve your problem.

michaelk · 09-14-2009, 06:22 PM

Assuming this computer has been working for awhile and no changes have been made to it I'm assuming according to the OPs first sentence I do not think it is a kernel issue.

You have posted incomplete snippets of information that really do not provide us with much information. Post the complete output of the lspci command in reference to the ethernet adapters. yada yada is not helpful. There is more then one Realtek chipset on the market. I assume it is a Realtek RTL8201 based on the sis900 reference. Do you also have a Intel PRO/1000 adapter installed in this computer?

My first guess is that the network adapter has failed. Have you tried rebooting the computer?

Louis_Carole · 09-16-2009, 02:00 PM

The node OS's will not be upgraded. They will die, and will not be replaced. We are already cannibalizing.

We rebooted (a bunch of times) as a first try. The network adapter did not respond.

Behavior for 5 years: up, used in parallel for computational chemistry
Behavior for the last couple months: down, neglected

Code:

# lspci | grep -i realtek
00:0e.0 Ethernet Controller: Realtek Semiconductor Co., Ltd. RTL-8139/8139C/8139C+ (rev 10)

There is no Intel PRO/1000 adapter on the computer. There are 8 lspci entries total, and none of the other 7 seemed relevant - I could be wrong.

I cannot cut and paste entire swathes, since there would be nowhere to send the data to. It does, however, force y'all to guess at what you don't know, which is unfair to you.

Let me know more specifically what info would be helpful to you. I may be able to find a floppy somewhere, if a full dump is what is required.

I am ready for bad news, and bad news might be why similar posts ended without resolution.

- Louis

TB0ne · 09-16-2009, 09:15 PM

Quote:

Originally Posted by Louis_Carole

The node OS's will not be upgraded. They will die, and will not be replaced. We are already cannibalizing.

We rebooted (a bunch of times) as a first try. The network adapter did not respond.

Behavior for 5 years: up, used in parallel for computational chemistry
Behavior for the last couple months: down, neglected

Code:

# lspci | grep -i realtek
00:0e.0 Ethernet Controller: Realtek Semiconductor Co., Ltd. RTL-8139/8139C/8139C+ (rev 10)

There is no Intel PRO/1000 adapter on the computer. There are 8 lspci entries total, and none of the other 7 seemed relevant - I could be wrong.

I cannot cut and paste entire swathes, since there would be nowhere to send the data to. It does, however, force y'all to guess at what you don't know, which is unfair to you.

Let me know more specifically what info would be helpful to you. I may be able to find a floppy somewhere, if a full dump is what is required.

I am ready for bad news, and bad news might be why similar posts ended without resolution.

- Louis

Well, have you tried to replace the NIC?? Could be that the NIC died, which is why the module(s) won't load and the interface won't come up.

michaelk · 09-16-2009, 09:50 PM

Yes it is unfair to try to guess with bad information.

The Realtek 8139 uses the 8139too module. If this is the only adapter then not sure why the modules.conf file contains a reference to the sis900 module. Maybe some hardware was swapped out at sometime but nothing was ever updated.

Look at the output of the dmesg to see if the adapter is being recognized. Look at the output of the lsmod command to see if the module is being loaded.

Try:
modprobe 8139too

If you do not see any messages then it worked. Edit the modules.conf file to replace the sis900 with the 8139too module.

You can then try ifup eth0.

Louis_Carole · 09-17-2009, 12:47 PM

I will try all these suggestions.

In the mean time, I do not know how people usually disseminate information about computers that are unable to connect to a network. Please let me know if there is a protocol for this - then I could give you the information you need, assuming I know what that information is

I just grabbed a bunch of bits from what I saw other posts were asking for, and am learning, e.g., the meaning of NIC, as I go along.

- Louis