No Network after power outage
 

Greetings,
I have a slackware 12.2 server. We had some really rough storms this weekend that took the power out for many hours. The UPS that it was connect to gave up the ghost and the server went down hard.

I powered up the unit and it some up fine but with no network. Ifconfig -a shows that it only knows about the lo interface. Both integrated gig ethernet ports are missing. I am not a slackware expert. Need to figure out what is needed to get it back on the network.

dmesg | grep eth0 shows nothing. Nor eth1. ifconfig eth0 returns no such device.

any help greatly appreciated.

does your ethernet device show up in lspci ?

This could well be a hardware issue. Since the power went down in a storm the network devices may have been hit by a surge. A lightning strike can induce a current in a network cable even without directly hitting it.

A couple years ago, I lost my onboard ethernet device in a storm (electric power surge). I had to swap in a replacement card to get it back up and running.

lspci -v output shows the ethernet ports.

what next ?

I am not an expert at bringing the eth0 interface up and down, but I do know that the script that controls the interfaces is

/etc/rc.d/rc.inet1

and it uses the configuration file

/etc/rc.d/rc.inet1.conf

Seeing that the hardware is still working, maybe simply issuing a

/etc/rc.d/rc.inet1 restart

would work?

Or perhaps the module is simply not loaded:

Does your lspci -v for the ethernet device state:

Kernel driver in use: r8169
Kernel modules: r8169

At least these are the modules that my computer uses to connect.

Tried the /etc/rc.d/rc.net1 restart. No output generated and no interface brought up. Modified rc.net1.conf to give verbose, but same result.

looked at dmesg output and saw NO ethernet interfaces. Did the lspci again and really looked at the output. both ethernet interfaces show up, but they have some ?. For example, the first one says:
01:00.0 Ehternet Conroller: Broadcom Corporation NetXtreme BCM5721 Gigabit ethernet PCI Exress (Rev 21). Subsytem: Dell known device 01eb
vital prduct data <?> Advanced error reporting <?> Virtual channel<?>
Power budgeting <?>

Both have the same ?'s. Does this mean my ethernet interfaces are fried ?

I think lspci reports what the card itself reports to the system. Noting that we see it all means at least the card has some communication skills left. And I think the <?> just means the card isn't reporting that information. And besides my card works, I'll show you what its output looks like:

It doesn't look like it shows kernel modules in use. Try
modprobe r8169 (At least that's the module my card uses)

Edit: The broadcom netXtreme should use the tg3 module
http://ja.broadcom.com/support/ether...aq_drivers.php

/sbin/modprobe -n -v tg3 generates the following message:
FATAL: module tg3 not found.

Looked around a bit and in /usr/src/linux-2.6.27.31/drivers/net found tg3.c and tg3.h.

The ethernet interfaces were working just fine prior to this outage. Maybe the module got corrupted ?

Hi,

Have you looked for /lib/modules/2.6.27.31-smp/kernel/drivers/net/tg3.ko or /lib/modules/2.6.27.31/kernel/drivers/net/tg3.ko depending on which kernel you are using?

Have you tried insmod?

Randy

/lib/modules/2.6.27.31-smp/kernel/drivers/net as well as
/lib/modules/2.6.27.31/kernel/drivers/net have tg3.ko.

When I modprobe I get an error. Haven tries insmod.

Hi,

I haven't used modprobe in a long time, but did you try /sbin/modprobe -n -v tg3.ko

Randy

If you need to reinstall the module packages, download (this looks like slackware 12.2 with the more recent kernel patch)

http://carroll.cac.psu.edu/pub/linux....31-i486-1.tgz

or

http://carroll.cac.psu.edu/pub/linux...smp-i686-2.tgz

and then as root
If the files were corrupt, this will replace them with a non corrupt version. If the files were fine, this of course won't do anything.

Solved!
 

Well,
When reading the modprobe man page for the fifth time, I noticed that modprobe would look in /lib/modules/'uname -r' for the modules. When I ran uname -r, I noticed that it returned 2.6.27.7 (I think, I am pulling this from my memory). Well when I looked in /boot, vmlinuz was a link to vmlinuz-generic-smp-2.6.27.31-smp. Since I only had this one boot image listed in lilo, I don't know how that happened.

I created a new link to the vmlinuz-generic-smp-2.6.27.31-smp. I create a new link to vmlinuz-huge-smp-2.6.27.31-smp. Modified lilo.conf and then did lilo.

Rebooted. Selected the first new link and it gave a panic. Rebooted and selected the second (huge) And the system booted fine. Ehternet interfaces up and running and configured.

The vmlinuz-generic-smp-2.6.27.31-smp must have been corrupted and an earlier image booted. Not sure how A 2.6.27.7 image was booted. But it is up an running and now I have some breathing room.

Have to say this has been pretty weird.