Red Hat 8.0, Kernel Version 2.4.29-pre2, ASUS P4P800SE MB, 3.2 GHz
Intel-P4 CPU, 1 GB RAM, ICH5R S-ATA only. SATA Disk = Maxtor 6Y160M0.

This is really busting my balls. Around August or September of last
year, a year and one month old (out of warranty) hard drive on my server
gave out and crashed the system. This was my newest at the time Maxtor
4R120L4 120GB disk which I installed as the server's root drive. It runs
really hot now after the crash, and it (I assume) has taken out 3
different brand new power supplies (230, 350 & 450 Watts). I bought a
new computer, this one, with two new sata disks and a 550 Watt power
supply, and copied the partitions of the bad disk (hdc) onto the second
sata disk (sdb) as giant 60GB files. I was able to run e2fsck on the
files and recovered most of my data by hand. Just about everything
turned up in the lost+found directories, and I was able to move a lot of
the stuff back to where it belonged. This took me most all of the
Christmas Holiday break. Since my grades at school suffered severely
during the quarter time period while I was trying to rebuild my server,
I promised my academic advisor I would not touch the computer again
until spring break. That is now, and my new sata drive (with all of my
recovered data and two weeks worth of Christmas Holiday work) seems to
have also fortuitously crashed just before I could transfer the data
back to my new server. ARRGGGHHHH!!!!

The setup is like this. I have a firewall with http and ftp software
that uses nfs to serve up files from this computer's brand new sdb1
partition. Not that it has anything to do with the problem, or was it
being used at the time, but one of those directories was writable as an
ftp "incoming" directory. All of the other directories were served as
read-only. The incoming directory was /linux2/ftp/incoming on this
system where sdb1 was mounted to the root directory as /linux2. The
recovered 60GB partitions were mounted as /linux2/hdcN.img to
/somerootdirectory using the loopback device. I haven't touched any part
of the loopback files since December and my recovery work. Except for
the /linux2/ftp/incoming directory having some occasional write accesses
to it from time to time, nothing has been going on with this drive until
I had time to transfer my recovered files this week.

This has been working very well, and without any problems since
December. A couple of weeks ago on March 6 I shut down the system to do
some school work on Win98, and when I rebooted linux, the system said
(incorrectly) that I had not unmounted my sdb1 partition cleanly and
needed to run fsck. I dutifully performed the fsck because most of the
time it never causes a serious problem. This time was an exception
because it found a bunch of errors, which I then elected to let it fix.

When I mounted the sdb1 partition afterwards, there were no files to be
found... not even the lost+found directory.

I have a feeling that Mator's SMARTs in this drive are not so smart
after all. Since this drive is still under warranty, it is no problem to
replace. I just need to know that I can still get the data back somehow.
E2fsed or whatever it was called.... isn't here anymore, and I guess
that doesn't really matter anyway because my 60GB files exceed the old
2GB e2fsed limit. "debugfs" says the following

[root@Sharlie bkup 20:05:27]# debugfs /dev/sdb1
debugfs 1.27 (8-Mar-2002)
/dev/sdb1: Attempt to read block from filesystem resulted in short read
while opening filesystem
debugfs: ls
ls: Filesystem not open
debugfs: open /dev/sdb1
/dev/sdb1: Attempt to read block from filesystem resulted in short read
while opening filesystem
debugfs: ls
...

I noticed that the last good bootup was around March 3, and after that
I started getting ata2 error messages. F-me! I left the G-dxxxxx "less
/var/log/messages.17" on the screen last night while the syslog rotated,
and now it's gone too. Anyway I have included as attachments in this
message the fsck error messages recorded. The system log error messages
can be found here at this web site
http://Christoffur.HopTo.Org/Photos/Uploads/

The main points of interest would probably be these files

http://Christoffur.HopTo.Org/Photos/Uploads/LastGoodBoot.txt 28530 Bytes
http://Christoffur.HopTo.Org/Photos/Uploads/FirstErrorMessages.txt
144961 Bytes
http://Christoffur.HopTo.Org/Photos/Uploads/Prtl-ErrorMessagesB4+AfterFsck.txt

48158 Bytes

Funny, there were no ata2 error messages on the 5th of March even though 
I performed two bootups. Unfortunately when the system booted at 15:57 
and I performed the two fscks, system logging does not seem to have been
enabled. There are no system log messages between 15:57 when I shut down
the computer to reboot and run fsck, and 17:01 when the system came back
up with the sdb1 entry removed from /etc/fstab. The above files have
been significantly trimmed to reduce the download size. If you want
access to the full and complete message logs, you can find them in the
Upload directory given above.

The problem with the message logs is that the messages from the new
sata ata2 drive look exactly the same as the messages from the old bad
hdc drive.

You can find those errors in the messages logs 99, 100 and 101 in the
same directory as above. It took three days to copy the data off of my
bad hdc drive onto the new drive.

I don't suspect an OS, or SATA driver problem at this time at all. It
would appear that the drive itself, while being only 3 or 4 months old
has developed some kind of hardware problem. I am not quite sure what
exactly the problem is, but it won't be until April before I can afford
another replacement disk.

I am going to power down and kick the cables on the bad drives and take
as many of the good drives out of this computer that I can, and put the
good ones back in the new 300MHz Pentium-3 server machine, which is
basically a September copy of this machine. That will probably take all
day, if I am lucky.

Hopefully 550 Watts should be enough power to drive the two bad disks,
and the one good sata root disk. Then I can hopefully copy the data off
of the bad drives onto the server were they belong. I have one week to
get this done and then my next quarter starts at school again, and I
really need to focus and concentrate on that. I am especially worried
about the root drive on this computer and the server because they are
exactly the same as the one that just crapped out. Except that the one
on the server is a regular ata drive and not a S-ata. The rest of the
drive is supposed to be exactly the same. If either one of those drives
blows up, I am really going to be sunk.

I am taking recommendations as to what type of drives to use to replace
my Maxtors. I thought that Maxtor was one of the most reliable drives
around and I have been using them exclusively for 10 years or more. It
would now appear that I may have been wrong about the Maxtor reliability
factor, or that something may have changed in their manufacturing
processes. I've got drives that have been running 24 hours a day 365
days a year for over 6 years without as much a hiccup. Now all of a
sudden in two years I've got two new drives giving me conniption fits
and driving me crazy.

I don't have the time nor can I afford the anger management classes
that would alleviate my serious ant-social behaviors that have developed
over the last few months related to these incidents. My poor defenseless
little 486 server, victim of the first hard drive crash, sitting
harmlessly on the floor awaiting a simple drive transplant, just
happened to get in the way of my big foot immediately following the last
crash on my new machine and suffered a fatal swift kick that caused a
compound processor neck bone fracture. Obviously this situation cannot
continue. If simple inanimate objects like my beloved little Shelby are
not safe from my personal and infuriating bouts of computer rage, I hate
to think of what the future will bing to the latest and greatest
pentium-4's that happen to lay near in proximity to my next coming
affliction.

If someone could please decrypt some of the error messages and let me
know if the 6Y16M0 data is salvageable and how that might be
accomplished, it might save me a couple of weeks worth of re-work to
re-recover my hdc data, and save future unsuspecting computer systems
from further days of unearned relentless hammering torment, retribution
and the rubbish heap.

Any other suggestions, observations or criticisms, as always, are
always welcomed.

Prayers should also be helpful, if anyone is so disposed, as this can
certainly not be an act of random chance, but instead due to the
critical nature of the timing of these events, this must be an act of
the hand of God. And I know not why? All I am trying to do is save the
world and the entire human race!

Regards from,
--
Christopher R. Thompson
Student of Mechanical Engineering
California State Polytechnic University, Pomona
home: http://www.csupomona.edu/~cthompson1/

SNAFU and .... Who was that guy who said.."If it can go wrong... it will
go wrong!"?