On Thu, Mar 27, 2003 at 11:38:14AM +0100, Tony Scholes
wrote:>
> Not sure whether this is an ext3 problem or not, but I've not seen this
on other RH systems.
>
> There is a particular file on an ext3 filesystem that hangs whichever
process tries to access it, i.e. an attempt to cp it will hang the cp. the
process is unkillable, and a reboot is the only way of getting rid of it.
>
> Is it possible that the file system is corrupted? If so how to get rid of
it?
Check the system logs, or use the dmesg program, and see if you have
any disk errors reported by the hard drive (ide or scsi) subsystem. A
bad block can cause the device driver to hang or repeatedly try to
read it (and failing, for a long time), thus replicating the symptoms
which you described.
> e2fsck -f finds nothing wrong.
If the bad block is in a data block, then e2fsck -f won't find it.
There is a chance that e2fsck -fcc (which will run badblocks in
non-destructive read-write test; see the man page) might be able to
find and fix the problem. It will take a long time, though, and
depending on the nature of the error, the device driver, the disk, and
the controller, might just cause the hang again, or it might
eventually get past the bad block if you wait long enough.
Good luck!
- Ted