On Tue, 2012-04-17 at 12:33 -0500, dimesio wrote: > I'm more inclined to suspect the hard drive based on your description > of the problem. The fact that reinstalling a file temporarily fixes > the problem would be consistent with a hard drive that is developing > more and more bad sectors over time. Both Western Digital and Seagate > have downloadable utilities that can be burned to a bootable cd to > check the health of their hard drives. > There are also a few standard ways to check a disk with built-in Linux utilities: smartd, fsck and badblocks. These should work with disks from any manufacturer. If the OP hasn't installed smartd, he should consider doing so as its a non-destructive way to check disk status and get warnings of impending problems. Once it is installed and running, giving the command: killall -USR1 smartd will cause smartd to do an immediate disk scan. If the disk that's suspected of having problems can be temporarily mounted on a PC as a second drive (an external USB disk dock is useful for this if the OP has or can borrow one) it can be checked more thoroughly by: - running fsck against the partition(s) containing /home and /usr with the -f and -n options set. The -f option forces a full scan and the -n option tells fsck to check the partition and report problems without attempting to repair them. - running badblocks against the disk device with the -n option Both programs should be used with the disk attached to the PC but without any mounted partitions on the disk, i.e. if the partition(s) get automounted they *MUST* be unmounted before running either fsck or badblocks. Martin
