Very Computer

I've installed mandrake 9 and everything ra fine so far but one day, the
machine hung (* but well, it does happen sometimes :)) so I manually
reset -> no way , "cannot find IDE drive" at bios analysis.

Ok, I power off, power on -> ok.
20mn later, the machine hangs again : the HD led stays on.

Power off etc .... : it works

Yesterday,I power on the machine and boot from my linux Hd : During the
boot process "Updating journal" : never ends and Led stays on.

I cannot boot anymore from this drive.

I'm planning to reinstall and destroy the faulty partition and recreate it.

Any better ideas ?

--
M?ns Rullg?rd

Vilmos

Backup my home dir to another disk.

Question : is there a tool to check the drive (mandrake 9) ?

thx for the answers !!

http://www.linux-ide.org/smart.html

Quote:> There are a couple of things that you can run, badblocks and smart suite,

> http://www.linux-ide.org/smart.html

    charly> Being Stuborn :), I let linux try to boot during the
    charly> update journal 10 mn later, he booted ok.

Did you  examine the logs (e.g.  /var/log/messages, /var/log/dmesg) to
see if the kernel choked because some hardware errors were detected?

    charly> Backup my home dir to another disk.

Good!  You can now safely play with that drive!  :D

--

Home page: http://www.informatik.uni-freiburg.de/~danlee

    charly> Greetings, I've installed mandrake 9 and everything ra
    charly> fine so far but one day, the machine hung (* but well,
    charly> it does happen sometimes :)) so I manually reset -> no way
    charly> , "cannot find IDE drive" at bios analysis.

I'd try to unplug the IDE cable, and replug.  Sometimes, this gets rid
of bad  contacts due  to rusts between  the contacting  metal surfaces
between the cable and the slot on the M/B or IDE adaptor.

    charly> Ok, I power off, power on -> ok.  20mn later, the machine
    charly> hangs again : the HD led stays on.

    charly> Power off etc .... : it works

    charly> Yesterday,I power on the machine and boot from my linux Hd
    charly> : During the boot process "Updating journal" : never ends
    charly> and Led stays on.

Indication of  a bad  drive.  After backing  up the valuable  data (if
any) from it,  don't use that drive for important  things any more, or
simply through the drive away.

    charly> I cannot boot anymore from this drive.

    charly> I'm planning to reinstall and destroy the faulty partition
    charly> and recreate it.

    charly> Any better ideas ?

My  "strange" experience  with a  bad drive:  it started  to  have bad
sectors.  So, I reformatted it  with 'mke2fs -c'.  While this prevents
ext2/3 to use the bad sectors for storage, I didn't feel safe, as more
sectors  may develop because  the hardware  is aging.   So, I  tried a
'mke2fs  -c -c'.   Wow!   It's a  miracle.   After that,  all the  bad
sectors are gone!  I can't believe  it.  So, I did 'mke2fs -c -c' once
more.  Again, no bad sectors.  The  HD LED didn't stay on for a minute
at   those  particular  sites   of  the   fsck  or   mke2fs  (actually
'badblocks').  I restored (unimportant) data  to it, and did some more
'fsck'.  No errors!

I'm still looking for an explanation.  :(

--

Home page: http://www.veryComputer.com/~danlee

    Andreas> [Bad sectors stay after read check but dissappear after
    Andreas> write check]

    >> Wow!  It's a miracle.

    Andreas> Not at all, just normal behaviour.

    >> I'm still looking for an explanation.  :(

    Andreas> Modern HDDs have a builtin defect management. Bad sectors
    Andreas> are automatically remapped until the drive is out of
    Andreas> reserve sectors.

I know.  So, when the drive  reported bad sectors, I assume it has run
out of  reserve sectors.   The remapping mechanism  is supposed  to be
transparent, isn't it?

    Andreas> Remapping of bad sectors on a read access would be very
    Andreas> dangerous so it is done on write access only.

So, it's not transparent anymore?

    Andreas> Some drives may also remap weak (i.e.  error correction
    Andreas> could retrieve the data) sectors on read access.

The other strange thing is  that the badblocks checking went smoothly,
with  no  noticeable  delays  when  touching  those  supposed-remapped
sectors.  No obvious  seeking noise is heard (which  is heard when bad
sectors are encountered).

So,  do  you mean  many  drives  can be  made  re-usable  (but not  as
reliable) by badblocks -n?

--

Home page: http://www.informatik.uni-freiburg.de/~danlee