On Tue, Aug 20, 2002 at 11:45:05PM -0400, James H. Cloos Jr. wrote: > First I got errors from the ide subsystem when trying to mount the fs; > rebooting from the suse 7.3 install cd into rescue mode I tried first > to mount then (after another reboot) to e2fsck(8). > > All e2fsck printed was that it was recovering the journal; all of the > errors were printk()s from ide indicating hardware errors reading the > relevant sectors. e2fsck was of course stuck in D state. > > The version on the rescue ramdisk is probably e2fsprogs-1.24a, as the > cds include e2fsprogs-1.24-11.src.rpm. > > The disk is failing and will be replaced, but I've been keeping it > going until that happens. If you're seeing that many errors, run, don't walk to the nearest computer store, buy yourself a new drivce, and transfer the data ASAP. Or do a backup right away, and pray. Disk errors very often have a tendency to cascade and increase exponentially, so "keeping a disk going" is often the very last thing you want to do. The other problem is that you're using an old e2fsprogs, which isn't as good about recovering from a bad journal situation. That's why you need to use debugfs to clear the journal flags.... newer e2fsprogs will do this automatically, although once you get to that situation you really are doing a last-ditch desperation "try to save what you can" triage situation. Many conservative sysadmins will replace a disk immediately as soon as they see soft errors; the moment you start seeing hard errors, you're generally better off replacing that disk in hours or minutes, if at all possible. In most cases the value of the data on the disk is far greater than the disk drive itself.... - Ted
