Hi,

I apolgize for my first posting being a troubleshooting question...

Anyway, today I was doing a little work in KDE 3.2 on Mandrake 10 and it all of a sudden froze. I couldn't get to a terminal, and the system would not respond to connection attempts via SSH. So I just hit the reset button hoping nothing would go wrong.

Upon boot up, it said there was a failure mounting /proc, which really concerned me. It then went into a forced filesystem check and found a lot of problems, resetting a lot of stuff back to 0 and then did an autoreboot.

The next time around it still said /proc failed, so fired up Disc 1 and tried the rescue options. I can't seem to be able to mount my ext3 volumes to run fsck on them. I even took a look at the drive with partition magic *cringe* and it said my volumes were corrupt when I "checked for errors".

Any suggestions on how to fix this problem? Or at least get reinstall the OS without having to reinstall and configure all of my software/settings? For the future, whould you recommend another filesystem besides ext3?

Thanks and Happy Fathers Day out there to all you dads!

~AIG

One possibility is that there was a hardware problem which caused the initial freeze-up. You may want to power down and re-seat the cables.

Are the drives excessively warm. They may be shutting down if over heated.

You run fsck on unmounted partitions.

/proc is a psuedo-file system, but it is mounted on root. Initially during bootup, the root drive is actually in memory. This is what the initrd file is. If the error occurred before the root directory is chrooted to your root drive, this doesn't sound good at all (like maybe memory problem).

The rescue mode has an lsparts command to list the partitions. Run that and see if the partitions show up.

Select the rescue menu option to mount the partitions on /mnt. If it is successful, cd to /mnt and look around to see if you are still in business.

You can change the root with the 'chroot /mnt' command after that.
Setting up the root $PATH variable and other variables can now be done by rerunning bash as a login shell. '/bin/bash -l'.

After that, if the /bin and /sbin and /usr/sbin directories exist, you should be able to run administrative commands normally. For example, you can run 'vim' and don't need to run 'vim-minimal' instead.

Load in the /etc/lilo.conf file. Make sure it looks ok.
Check the /etc/fstab file.

You can now try mounting the rest of the partitions with the 'mount -a' command.

Re-running lilo at this point might be a good idea. If it fails, you will need to correct the /etc/lilo.conf file or resolve the problem with the partition.

Good Luck! Someone may respond with a more particular answer about the particular error message you received.

although i've never tried it, i hear mounting a drive before fsck can trash it. just do a 'fsck /dev/sda1' or similar.

i know a few people that have trashed their drives with partition magic and would personally avoid it.

as for ext3 it's a good filesystem in my experience (have many drives happily spinning with ext3 for long times).

Thanks for the responses. I was able to recover the rest of my data and fix the filesystem, but mouting /proc was still a problem. I did an in-place upgrade to see if that would help. There were enough little things wrong that I decided to just do a reinstall, which I didn't mind.

Thanks for the help.

~AIG