Looking to linux for hardware troubleshooting
 

I have a computer with two network cards (one onboard and the other PCI). After shutting it down and turning it back on, I now I can't get the network cards to work. I ran the latest Knoppix Live CD (5.01) and no network interfaces are recognized. I also installed SuSE 10.2 and no network interfaces were recognized.

The network cards were recognized previously under Clarkconnect (CentOS). Is there anything else I can try with linux to confirm or deny that this is a hardware problem?

mmm thats weird.. If the cards work in one OS then it shouldn't be a hardware problem. Try to configure them in SuSE with NEtworkManager. It worked well for me.

Could be that the kernel modules for the network cards aren't being loaded.
Do 'lsmod' to see if the proper modules are being loaded.

If not then do 'modprobe <modulename>' to try and load them. If it says module is not present then you'll need to add support for it in the kernel.

If you don't know which modules you need, then post the output of 'lspci' and I or someone else can try to help you.

I checked lsmod and couldn't see anything related to network cards. I would like to post lscpi, but I have to save its contents to a USB flash drive and am having trouble mounting it. I plugged the USB drive in and the HD light flashed for a few seconds so I know that the system knows about it, however I do not know what device it is to mount it. How can I tell?

Do 'ls /dev/sd*'

It'll be one of those devices, e.g. /dev/sda1

Whether it is sda1, sdb1, sdc1, etc depends on your system setup, such as if you have other usb storage devices connected or if you have SCSI hard drives.

If you know the name of the relevant driver(s), you could do modprobe [driver].

Okay, I found my USB drive. It was /dev/sdb1

The following is the output of my lspci:
I don't see anything regarding the motherboard's onboard LAN (eth0) and my add-on PCI network card (eth1)???

Out of curiosity, do you remember which kernel version you were using in Clarkconnect and which kernel version you're using now?

Also, are you in Suse or Knoppix?

You are only showing an onboard video (as an unknown device, yet), and none of those VIA devices should be unknown.

Looks from here like you could have had a hardware failure.

From your wording, I infer that the network devices WERE recognized under CentOS, then they WERE NOT recognized under CentOS, then you installed SUSE and they were not recognized and they also are not recognized under Knoppix? Is that correct?

You might try doing a modprobe with the specific drivers. More than likely the Via onboard NIC needs the via-rhine driver (or a variant); might try just stuffing it in. It'll probably go even if some of the device startup reporting is down.

In any case, if my interpretation of events here is correct (it worked, then it didn't on the same OS), I'd bet you're looking at a new motherboard.

Personally, I steer clear of these all in one boards, and now you see one of the major reasons why. Give me slots, let me plug in cards to implement features. Also, let me change cards in slots to upgrade features or to fix problems.

If this is the case then I agree.
I just don't want to come to a conclusion without knowing as much as I can about the situation as I have seen quite a few strange things happen when working on machines.

Absolutely. Hence my caveats.

Gentlemen, you are correct. Thanks for your help. It worked under Clarkconnect (CentOS variant) then one day stopped working. No network cards were detected subsequently with a new install of Clarkconnect, SusE 10.2 and running Knoppix 5.01 (as a live CD). Ergo, I suspected a hardware failure. Today I took it into the computer store where I bought the motherboard and processor. When I turned the machine on and the tech people saw linux, their eyes glazed over and I got the "Gee .... linux.... we only support Windows" routine. Okay fine, just test the mb and network cards then. I'm sure that they will say the motherboard (which was an Asus M2V-TVM) will need to be replaced.

Edit ---> BTW, if the motherboard needs to be replaced, can anyone suggest a better mb for linux suitable for a AMD Sempron processor?

Asus makes a fine board, and presuming your board has failed the symptoms say it has had a silicon failure (not a filter cap, resistor, or power supply related failure).

Silicon failure almost certainly means one of two things: thermal problems or power spike. Third possibility is abuse: incorrect voltage settings, or improper overclocking - some such.

Power spike *could* mean faulty power supply, but it could also mean your input surge suppression is insufficient.

If yours is a cheap no-name power supply, without lots of acceptance stickers from lots of ratings organizations (such as underwriter's labs backward R symbol) then when you replace the mobo you should also ditch the PS.

If the system is connected to a power system that is unreliable, then you should at a minimum replace the surge protector and you should give some thought to purchasing an uninterruptable power supply. If your electrical system is old enough that it is ungrounded, then you really need to provide a good ground connection for your 'puter.

Of course, if you have thermal problems, you need to solve them.

Probably the failure is in the southbridge chip. On some systems that chip gets hot and not all boards put heatsinks on them.

Losing hardware because of any kind of environmental problem is a baaad thing. Don't just change the mobo; correct ANY deficiencies that might have been responsible for its death.

JMHO, of course.

Thanks for the good info. The power supply should be sufficient as it was part of an Antec NSK1300 micro ATX case. Heat will be one thing I'll look at though, especially in the small case.

Much as I like Antec cases (my workstation resides in an Antec P-182), I have to say that I know that particular case to run hot. Also, that particular mobo seems to cram the components in.

I think it is a bad idea to try to shoehorn a high performance system (and a 64 bit system IS a high performance system) into a tiny case. Thermal management becomes a very serious issue in that event.

I used to do that sort of thing, designing for the military, and let me tell you that we had some really serious issues getting the heat out of avionics boxes that were stuffed full of electronics.

I would suggest that you get a bigger box and make sure the cooling capability is substantial.

Go to tigerdirect.com

You can usually find some good cases there that don't cost too much.