[plug] root filesystem problem on Debian testing/unstable system

[Brad Campbell]

Bret Busby wrote:

Glagd you got it all sorted Bret.

> The lessons from this (I think), are:
> 1. do not run fsck on a mounted partition (the root partition was mounted when I ran fsck to check it and everything went awry), and
> 2. when problems arise with a root partition (that is ext3), in terms of it becoming an unrecognisable, unmountable filesystem, with stated bad FAT blocks, a good idea is to run e2fsck on the root partition, booting from a Knoppox 4.0, or similar, CD/DVD.

Yes yes yes yes.. fsck should have complained _loudly_ if you asked it to run on a mounted fs if it 
was r/w. And yes it's always a good idea on a broken filesystem to check it unmounted completely.

going back to your original post I spotted this...

"I booted the computer, using a Knoppix 3.1 CD, and tried to mount the root partition, to run fsck 
on it, to fix it, but the partition could not be mounted, as the filesystem was unrecognised. Using 
Knoppix, I was able to view the contents of other partistions on both the physical hard drive that 
conatins the root partition, and another hard drive in the same computer"

Never try and mount an ext3 partition prior to running fsck on it if you think it may be broken. 
fsck does a much better job of cleaning up a broken filesystem than the kernel will when it replays 
the journal as you try to mount it.

I'd love to know what killed your fs though. I've been running ext3 on all my machines since it hit 
the stable kernel, and I've had exactly what you saw before, but on mine it was caused by an 
intermittent ide connection inside my laptop. I've _never_ seen anything like that on a normal system.

Might be a good time to fire up debsums and make sure all your OS files are free of corruption.

Brad
-- 
"Human beings, who are almost unique in having the ability
to learn from the experience of others, are also remarkable
for their apparent disinclination to do so." -- Douglas Adams