[wplug] FileSystem problems

[Vanco, Donald]

Mike Griffin <mailto:mike at dmrnetworks.com> wrote:
> I seem to be having some problems with one of my servers and was
> wondering where to start troubleshooting the machine. I would imagine
> that it's at the hardware level.
> 
> A fileserver crashed yesterday with a kernel panic. This machine has
> been running for nearly a year, solid. I started getting errors on the
> hardrive during writes to the drive, and accessing was very slow. I
> ran maxtor utilites on the drive and it found a few problems that the
> software fixed. I reinstalled the server OS (RH7.3) and performed my
> data recovery from backups. My backups are stored on the same system
> but on a different HDD, which gets mounted with a script everynight
> and has tarballs written to it, I also ran fsck -t ext3 on this drive
> (/dev/hdb1).  I checked for a new backup this morning, and all was
> well. I just tried mounting the drive a few minutes ago and had a ton
> of bad sector attempt timeouts saying it cannot find a valid FAT
> partition. I tried to run fsck on this partition I get this as a
> result: 
> 
> [root at fileserver root]# fsck /dev/hdb1
> fsck 1.27 (8-Mar-2002)
> e2fsck 1.27 (8-Mar-2002)
> fsck.ext2: Attempt to read block from filesystem resulted in short
> read while trying to open /dev/hdb1
> Could this be a zero-length partition?
> [root at fileserver root]# fsck -t ext3 /dev/hdb1
> fsck 1.27 (8-Mar-2002)
> e2fsck 1.27 (8-Mar-2002)
> fsck.ext3: Attempt to read block from filesystem resulted in short
> read while trying to open /dev/hdb1
> Could this be a zero-length partition?
> 
> These are two different ATA drives. One is a 20G and one is a 10G. I
> thought it was kind of weird that this would happen to both drives one
> day apart. possibly a controller problem on the motherboard?

	Are all your cables (power and data) in good shape and well seated?
	Any chance there's been a change in quality of the power?
	Do BIOS and kernel see the drive geometry in like fashion? (there
was a time, circa RH6.1 or .2, that fdisk added an "extra" cylinder to the
drive - fun!)
	Do you have a "like system" you can swap the drives into to see how
they behave?

	IMHO - using IDE HDD tools to "fix" a drive that's reporting media
errors is like putting a band-aid on a leper.  Underneath, it's still
rotten.  If you've got a drive spitting bad sector errors it's time to
$h!tcan the drive and get a new one... MaxTools may delay death, but it's
still terminal, and data loss is almost assured.

Don