13 Nov
1998
13 Nov
'98
12:36 p.m.
About the middle of the row of racks a red fault light
flickered, then came
on brightly. The drawer of the middle disk moved out by itself and when the
top opened a greyish black cloud came out, paused for a moment over the
drive, and then was whisked away by the air conditioning.
Wooha - it opend automaticly ? Dangerous. The first thing I
learned on head crash was:
Stop the unit
Unplug power
Unplug Interface
Roll the unit out of the computer room
At best out of the house
Now have fun with the damaged parts :)
At all this steps (but the last), especialy when
a delay ocures, guard the top with your life.
We once had a 144 MB unit crashing on saturday
morning and running (the check logic was damaged)
until monday morning (no operators). The heads have
been scraped completly, and the crashed disk was just
pure alu - no magnetic surface left ... The fun thing
was that parts of the disk stack had been readable until
sunday night ... We couldn't belive the transaction
protocolls :)
"I believe that pack has crossed over to the
other side." said the SP in a
complete dead pan protestant minister delivery.
:)))))
Gruss
hans
--
Ich denke, also bin ich, also gut
HRK
13 Nov
13 Nov
12:38 p.m.
New subject: Favorite Head Crash | reply somewhat Off Topic
Reading this reminds me of a "notesfile" we used to have within DEC
called "warstory" which was a "compilation" of misadventures dealing
with computers. I wonder if the new or next incarnation of the list
could have something similar as a subset.. I'll bet there's _loads_
of humorous stories of misadventures involving classic equipment..
At DEC, we had loads of them.. Hey Allison, I bet you've got a few!
Will
Hans Franke wrote:
About the middle of the row of racks a red fault
light flickered, then came
on brightly. The drawer of the middle disk moved out by itself and when the
top opened a greyish black cloud came out, paused for a moment over the
drive, and then was whisked away by the air conditioning.
Wooha - it opend automaticly ? Dangerous. The first thing I
learned on head crash was:
Stop the unit
Unplug power
Unplug Interface
Roll the unit out of the computer room
At best out of the house
Now have fun with the damaged parts :)
At all this steps (but the last), especialy when
a delay ocures, guard the top with your life.
We once had a 144 MB unit crashing on saturday
morning and running (the check logic was damaged)
until monday morning (no operators). The heads have
been scraped completly, and the crashed disk was just
pure alu - no magnetic surface left ... The fun thing
was that parts of the disk stack had been readable until
sunday night ... We couldn't belive the transaction
protocolls :)
"I believe that pack has crossed over to the
other side." said the SP in a
complete dead pan protestant minister delivery.
:)))))
Gruss
hans
--
Ich denke, also bin ich, also gut
HRK 
16 Nov
16 Nov
1:12 a.m.
When I was in college, interning with the Social
Security Admin in 1982, I visited a huge disk farm.
Easily 200 dishwasher sized floor standing units in
a vast array stretching across a large room. As my
guide was giving me his canned talk, I saw something
strange out of the corner of my eye. I interrupted
him and he told me to just keep scanning across the
units. Every minute or so, one of the drives would
puff out a small orange cloud. I was told that each
puff was a head crash. The drives would expel the
orange oxide that had been scraped from the surface,
the head would almost always be undamaged and the
drive would keep right on going. The bad blocks
would be noticed and marked by the OS and when a pack
had enough bad blocks it would be pulled and sent out
to be resurfaced.
4:53 a.m.
A non-text attachment was scrubbed...
Name: not available
Type: text
Size: 1622 bytes
Desc: not available
Url :
http://www.classiccmp.org/pipermail/cctalk/attachments/19981116/ac022ecc/at…
6:55 a.m.
units. Every
minute or so, one of the drives would
puff out a small orange cloud. I was told that each
puff was a head crash. The drives would expel the
Not a real headcrash (because IMHO that implies the heads don't take off
again), but more like HDI (Head-Disk interference). This can, indeed, be
non-fatal - in fact some hard disks are designed so the heads hit the
surface when they load and then fly from there. It always worried me, but
it seems to be OK.
On the other hand, if each drive experiences HDI at 3 hour intervals (on
average), then I would suggest there's something wrong. Possibly they are
in need of filter changes or something. orange oxide that had been scraped from the
surface,
the head would almost always be undamaged and the
drive would keep right on going. The bad blocks
would be noticed and marked by the OS and when a pack
had enough bad blocks it would be pulled and sent out
to be resurfaced.
Minor HDI doesn't even cause bad blocks...
But the idea that bad blocks are continuously being created and then
ignored by the OS also worries me. What if there's some important data
there that needs to be read back. Of is this a WORN (Write Once, Read
Never) system :-)
9536
days inactive
9539
days old
4 comments
4 participants
participants (4)
-
ard＠p850ug1.demon.co.uk -
bill＠chipware.com -
franke＠sbs.de -
wpe101＠banet.net