[Gllug] Dead Reiser drive

Peter Grandi pg_gllug at gllug.for.sabi.co.UK
Fri Oct 21 15:20:47 UTC 2005


>>> On Fri, 21 Oct 2005 12:30:49 +0000, SteveC
>>> <steve at asklater.com> said:

steve> A friend asked me to look at his 120 gig reiser drive
steve> that wouldn't mount. If I try to dd or badblocks it then
steve> I get some errors

[insert humorous/irreverent comment here about vagueness] and
what are these errors like?

steve> and then it locks up IDE and I can't even shutdown -r
steve> now, it hangs trying to sync.

steve> So I think it's gone to that big IDE controller in the
steve> sky,

Good guess :-).

steve> unless anyone has suggestions other than paying someone
steve> in a cleanroom to do some magic?

Well, according to rumours, the guys in the cleanroom don't
necessarily do much; such vicious gossip as I have read/heard
suggests that most recovery companies do perfunctory activities,
like running Norton Unformat. The military and intelligence
agencies may have platter recovery scanners, but apparently most
commercial recovery companies stop well before that.

However there are the usual semi-reliable-but-plausible recovery
attempts one can do that are:

* The usual mythological ''put the drive in a fridge overnight'',
  this may cool down the electronics enough that if it is a
  failed IC it may work long enough to do something (cfr. why
  there are ''freeze sprays'' on the catalogs of electronics
  supplies companies).

* Shake the drive a bit hard while not operating and try again;
  the drive a bit less hard when operating and try again. The
  shakes may help if there is sticktion or similar, and may
  damage some areas of the drive, but something better than
  nothing.

* Find an identical drive and swap the controller board, if it
  is the controller board that is defective. Allegedly here are
  per drive calibration data on NVRAM in some boards, so who
  knows.

* Open the drive and check whether there are some mechanical
  problems e.g. by manually rotating the spindle or twisting the
  arm in case there is something like sticktion. The cleanroom
  bit is ideal, but not necessary, as long as one takes _some_
  minimal precautions (e.g. gloves to prevent skin particles
  from being shed, not using the spindle area as an ashtray,
  ...) and is willing to accept some risk. I have seen drives
  that have been opened even without many precautions continue
  to work, at least long enough to get stuff off them.

* Conceivably it _might_ work to transfer the spindle from a
  broken drive to a new one, but I suspect that would require
  extraordinary care (the spindle is heavy and spins 120 times
  per second) that only a lab or a manufacturer would have the
  equipment for.

ObDisclaimer for the ''snipe maniacs'': nothing of the above is
guaranteed to work, and indeed may have a low chance of working,
it is mostly just random folklore, but what you got to lose?

-- 
Gllug mailing list  -  Gllug at gllug.org.uk
http://lists.gllug.org.uk/mailman/listinfo/gllug




More information about the GLLUG mailing list