"ifup eth0" fails at boot

[nemolomen]

I'm having a weird problem.

Lat weekend I started having problems with my HDA in that I would lose all DNS, DHCP, etc... on the network about 5 minutes after rebooting the server.

This morning I was able to note the following by entering interactive startup mode:

When starting the network interface I see...

Usage: grep [OPTION]... PATTERN [FILE]...
Try 'grep --help' for more information.
Usage: grep [OPTION]... PATTERN [FILE]...
Try 'grep --help' for more information.
Cannot find device "up"
Failed to bring up .

I can also get the same output when entering "ifup eth0" at the terminal when logged in as root .

Has anyone encountered anything like this before?

Thanks in advance,

nemolomen

[cpg]

this smells like some kind of hardware change or corruption in the configuration files in /etc/sysconfig/network-scripts.

the error seems to be traced to this system script: /etc/sysconfig/network-scripts/ifup-eth

Code: Select all

bash$ grep 'Failed to bring up' /etc/sysconfig/network-scripts/ifup-eth echo $"Failed to bring up ${DEVICE}." bash$

however, i could not debug much more. as you can see, DEVICE is empty by the time the code gets there. so, somewhere between /sbin/ifup and ifup-eth, that variable is lost. given that this was working, it seems odd that it would fail. perhaps some sort of hardware change? or a driver update caused the device to be gone? or (gulp) file corruption?

[nemolomen]

Yeah, um, I'll take file corruption for $1000 Alex.

I managed to get almost as far with tracing the scripts.

The hardware has not changed since a installed the gigabit network card a few months ago and I haven't done any manual updates in at least a month.

I did discover one of the files in /var/logs was 8.8 GB. It filled the drive and I was getting disk full errors. I was able to delete it so I could actually do other stuff.

Guess I'll have to start over from scratch with a new install, that is unless the backup of /etc I have will do me any good. I haven't tried restoring it yet since I wasn't quite sure if it could make things worse. I've been saving that as a last ditch option. But it sounds like I should also have a backup of /dev, which I don't.