[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
InitTab, Disk Drive Dust... (fwd)
pretty interesting walk-thru on drive death signs and other aspects. not
directly Linux specific, but interesting nonetheless.
---------- Forwarded message ----------
From: Javilk <javilk@polly.mall-net.com>
Cc: svlug@svlug.org
Subject: InitTab, Disk Drive Dust...
...
> > 12:123:respawn:/sbin/mingetty tty12
> > And reboot.
>
> Reboot? REBOOT? What, is this Microsoft UNIX for Windows??
> Don't reboot! Just send init (PID 1) signal 1!
>
> kill -1 1
You mean I been rebootin fer nutin???
Well, yes one can do that, but rebooting has always insured that
everything is properly reset from my latest ossified screwup, and reveals
other "conflaguration" file screwups before I forget what the heck I've
been screwing with in the moonlight.
> You rarely have to reboot a unix machine. Kernel upgrades,
> yes. Major software revisions (rh4.x -> rh 5.x), yes.
> Editing /etc/inittab -- NO! Installing a web server -- NO!
> Setting up access control lists -- NO! Modifying sendmail.cf -- NO!
> changing IP addresses -- NO!
And I will probably remove a spare hard drive from this machine without
rebooting. umount ... pull the power plug, and then the SCSI plug. Don't
do it the other way (adding a drive,) as if you get the plug polarity
wrong, you might still contact the +12 to the +5 and kill the drive. I
did that once... Not because I was trying to keep the machine running, but
because the other hard drive had a serious start up problem, and it was
worth the price of a drive to try to salvage that drive.
A little bit of disk lore... Sometimes, hard drives become hard to
start, long before they fail. As drives age, the more volatile components
of the oils evaporate. This can result in harder starting torque. Also,
some of the power transistors might die or weaken, reducing the start up
force. Often, a gentle tangential tap at the right instant will help them
spin up again. That tap can become harder and harder with age. In this
particular case, it eventually meant removal of the drive, and whacking
till there was blood on my hand the last few times I spun that drive up.
(We Really, Really needed the data! And it had always responded well to
the occasional tap in the summers, but not winters. The machine was only
powered down two or three times a year on average, mostly due to power
failures.)
Well... after adding another drive, and copying the data off, I
removed the drive cover to observe the spin-up sequence. It seems I had a
lot of my understanding of the noises wrong. And it seems, gotten my hand
bloody for nothing!
When you hear the initial flutter of a recalcitrant disk drive
starting up, it is not the heads swishing back and forth; but the platter
assembly hesitating back and forth; the platter assembly not spun forward
enough to engage the magnetic forces of the next set of coils. So the
next set of coils moves the wrong magnetic poll backwards, instead of
moving the right magnet poll forwards.
Holding the drive by fingers above and below the spindle axis, a very
gentle tap in the appropriate direction (try both) DURING the flutter has
a very good change of helping the platter assembly "catch" the next coil,
and start up. The multiple flutters are repeated attempts by the logic to
start the drive. Given what I observed, I rather think that one or more
of the coils may be blown, and the drive still started.
Other drives may start up ok, but stop after running a short while
due to problems in reading the media. I've seen this behavior in a few
drives. Some will then blink their activity lights in a tell-tale
diagnostic pattern.
Much of the "whir" you hear, is the incredible movement of air inside
the case. There was a patented turbine that used flat plates instead of
blades, and examining a spinning drive, one can see why! As the drive
spins up, it forces a little vane off to the side, unlocking the heads.
This is the little click you hear in modern drives. Some of the older
voice coil drives have a more pronounced click due to a solenoid that
unlocks the heads.
Some people speak of "stiction problems", as if the heads are
sticking to the platters. I did see this once on a drive that I
dismantled, but long before the platters are held immobile by head
stiction, one is likely to find tons of bad sectors on the drive due to
the oils. I saw that on an 80 meg Apple drive that I picked up at the
flea market for a few dollars once. It too, had to be hand started, and
revealed a lot of bad sectors. Examination revealed oil spots on the disk
surface, spots easily wiped off.
Of course, opening the drive in anything other than a clean room will
tend to ruin it. However, I wonder just to what degree this is true, given
modern plated media...
Most drives have an air filter somewhere to capture any lose debris,
and often some adhesive surfaces to help. It is possible that modern
drives are far more resistant to continued degradation due to particulate
damage than most people suspect.
Being somewhat of the hands on mad scientist type...
Experimenting with a more recent plated media drive with other
serious problem, I opened an inspection port on the side of the drive, and
poured some silica gel particles to see how much garbage it would take to
stop the drive, or at least render it difficult to start, and what kind of
sounds it would emit. (I had had another drive that made sounds like a
orbital sander, and wanted to see if it was head debris, or bearing
problems that had caused those loud awful clattering noises.) Modest
amounts of silica particles did not stop the drive, nor interfere with
starting characteristics, as the particles were flung outward and caught
by the adhesive surfaces and the air filtration system inside the drive.
And of course, flung back out the inspection hole with some vigor! On
disassembly, surface damage to the media was surprisingly light, just a
few visible scratches. My conclusion was that the other drive had had a
catastrophic bearing failure, probably the fragmenting of several ball
bearings or roller bearings.
I currently have a Conner CFP2105S with a chip that becomes
exceedingly hot. The media is probably ok. Anyone have a dead CFP2105S we
can try to swap boards on?
> > # Note: Do not use tty7 in runlevel 3, this virtual line
> > # is occupied by the programm xdm.
>
> It's actually occupied by /usr/X11/bin/X (or whatever X is symlinked
> to.) I thought X picked the last free device, so if 7 was in use,
> it would use 8, if 8 was in use, it would use 9, etc.
Well, last time I fudged with this one, it caused respawning errors.
That was Caldera's 1. something a few years ago, so it may be better now.
- javilk@mall-net.com -----------------------------
-------- MS asks "Where do you want to go?" -------
------- Linux asks "What do you want to do?" ------
-- It is doers, not goers, who built this world! --
--------- Member: http://www.svlug.org/ -----------
--
echo "unsubscribe svlug" | mail majordomo@svlug.org
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ to unsubscribe