Re: Serious problem with Linux on an old PC

In message <44f474d7$0$22495$626a54ce@xxxxxxxxxxxx>
Bernard <debreil@xxxxxxx> wrote:

On Tue, 29 Aug 2006 15:46:50 +0200, The Natural Philosopher wrote:

Bernard wrote:
On Tue, 29 Aug 2006 13:20:05 +0200, The Natural Philosopher wrote:

Bernard wrote:
Hi there !

Could anyone give me a diagnosis on my hard disk problem ?

On an old Fujitsu, I had installed RedHat 6.0, a few years ago. It
worked for years
without any problem. Then, last june, I had a crash on a partition,
and I had to re-install, which I did only last week. At first,
everything seemed OK, but, after 2-3 days and about 3 ou 4 shutdown
and startup, I ran into a problem similar to that of last june,
except that it occured in /dev/hda5 instead of /dev/hdc3 last time :

'/dev/hda5 contains a filesystem with errors ! check forced...

... kernel panic, [file system] inconsistency... run fsck
manually...'

at this point, I typed the root pasword as required, got dropped to a
shell, and typed:

'fsck /dev/hda5' :

'PASS 1 : Checking inodes, blocks and sizes... Duplicate blocks found
..... Invoking duplicate block passes... etc...

PASS 1B : Rescan for duplicate bad blocks... Duplicate bad blocks in
inode 55299 : 221452
55300 : 221453
55301 : 221454
... etc...

PASS 1C : Scan directories for inodes with dup blocks ..............
PASS 1D : Reconciling duplicate bad blocks

file ... (inode 55305, mod time Sat Aug26,2006 has 3 duplicate blocks
shared with 3 files.....

etc.., etc..,
/etc/mail...

Clone Duplicate bad blocks => Y

Pass 2 : checking directory structure : the '..' in /etc/mail (55299)
is missing... FIX => Y

PASS 4 : checking reference counts: inode 2 reference count is 15,
should be 16 FIX => Y, inode 18433 reference count is 27, should be
26 FIX => Y... etc... unattached inode 55301

etc...

PASS 5 : checking group summary information. Block bitmap differences
+221461 +221462 .... 221486... etc...

etc... etc... it took at least 15 minutes to fix everything, replying
'Y' to every ask. Then I got :

'/dev/hda5 : filesystem was modified. /dev/hda5 6703/110592 files
(0.4% non contiguous.

on reboot, I had:

/dev/hda5 clean
/dev/hdc2 clean
/dev/hdc3 clean

After so many fixings, I rather expected that it would not reboot
properly, but, so far, I have not found anything wrong in the
repaired system.

Last june, about the same thing had happened, not on /dev/hda5, but
on /dev/hdc3. There were so much fixing to be done, that I had given
up. This time, it happens on a newly re-installed system.

I suppose that a similar problem is going to happen again some time
sooner or later...

Thanks in advance for any useful input
Essentially it looks like the disk is falling apart.

Get a new disk, and reinstall on that.

Yes, but then, why did this happen last june on /dev/hdc3 and now in
/dev/hda5 ? These are two different disks (my PC has 3, the third one
being hdb). It seems curious that, after 8 years of good service, both
disks would give up at the same time or just about... I have heard of
PC lasting 15 years or more...

Mmm. We generally found SCSI disks gave up about 5 years in..

But don't discount the possibility that something - dust, temperature,
or even a power failure - has damaged both at the same point.

However another possibility - that of a failing SCSI controller -
exists..

But these are not SCSI disks, just IDE disks


The answer there is to borrow someone else's SCSI machine and try and
mount and fsck the drives on that.

I have to say after many years of trying to be smart with disks, I
simply bin them at the first sign of trouble.

If I were sure that those disks are faulty, I would replace them...
provided that I can find such old standard disks for such an old
machine...

Replace them!

Retail IDE disks seem to have a life of around 3 years. (Older ones,
e.g. <1GByte lasted longer, because when they were made they were
expensive, premium items.)

SCSI disks generally last longer, but I don't have the experience to
suggest how much.

About 4 years ago there were a considerable number of 20GB disks with
a manufactirung fault. The majority of them failed between 2 and 3
years old. We had 15 of them. 1 failed in its first month. 3 more
failed before 2 years. 8 more (including one of the replacements)
failed during the 3rd year. The vendor eventually swapped the lot.

Failure clusters can, and do, happen.

Alan


--
Alan Adams, from Northamptonshire
alan.adams@xxxxxxxxxxxxxxxxxxxxxxxxxxx
http://www.nckc.org.uk/
.

Relevant Pages

  • Re: Final MRL#1 DVDs+CDs!
    ... master DVDs for me, in quantities of less than 1000 units - and the ... The only companies that duplicate disks in quantities that small, ... use the same type of disks as Luke is using for this ... DVD in small quantities at reasonable prices - I would seriously ...
    (rec.juggling)
  • Re: MYDOS 128K
    ... disks to duplicate. ... versions that magically used extended memory without regard to ... I have used an extended memory sector copier lately, and this one was written in 1988: the "SECTORCOPY 1.5" off the Speedy System Disk ...
    (comp.sys.atari.8bit)
  • Re: Serious problem with Linux on an old PC
    ... Rescan for duplicate bad blocks... ... /dev/hda5 clean ... We generally found SCSI disks gave up about 5 years in.. ... IDE drives don;t generally do more than 3-5 years at best. ...
    (comp.os.linux.setup)
  • Re: Serious problem with Linux on an old PC
    ... Rescan for duplicate bad blocks... ... /dev/hda5 clean ... We generally found SCSI disks gave up about 5 years in.. ...
    (comp.os.linux.setup)
  • Re: JDS dust collector
    ... Hardly any healthy pins look Winifred, and they monthly excuse Jimmie too. ... Better clean disks now or Stephanie will regularly climb them against you. ...
    (rec.woodworking)