Print

Print


=============================
From:    Jon Lewis <[log in to unmask]>

> Yes, I could simply buy a machine (and might do so soon anyway), but
I
> hate to scrap perfectly good computers, especially given how well
they
> run Linux.

I've run 160 and 250GB drives in what some would consider some really
old 
clunkers (PII-333 using Promise Ultra100TX2 since the built-in IDE was

only ATA-33, and I currently have two old Tyan PIII (~1Ghz) systems 
running pairs of 160s using the built-in Promise controllers (but not 
Promise's softRAID).  In those, I did need a Promise BIOS update to get

proper support for drives larger than X...can't remember if it was
128GB 
or 137GB.

If a BIOS update won't solve your problem, you might look into getting
an 
IDE (or SATA) interface card, which will likely solve the problem.  If

your machine is old enough, the IDE interfaces may suck anyway.  If
they 
are ATA-100, I'd be surprised if a BIOS update wouldn't be all you
need.

You really want one interface per drive though, so if you want 3 500s,
I'd 
suggest a 4 (or more) port SATA card and SATA drives.
=============================

I would be pretty happy with one bigger ATA drive.  Basically, I have a
lot of older but still very nice hardware and would like to cobble it
together into something I can use for a while.  For various reasons, I
have one box that is essentially an in-house internet cafe, and I am
leaning toward using Ubuntu for that purpose.




=============================
From:    Shawn McMahon <[log in to unmask]>
> haywire when it writes too far some day?  I have read some
statements
> that Linux does not use the bios and is therefore immune to any such
> problems, but I find that hard to believe.  Somewhat more sensible
was a
> suggestion to disable on-board controllers in favor of an add-on
> controller that knows what to make of the drives.  Would such a card
> auto-detect parameters, and/or include its own BIOS?  Is there any
> reason not to do something like that?  My main concern is to be able
to
> later put the drive in another machine w/o worrying about special
> drivers.

Linux doesn't use the BIOS and is therefore immune to that particular
problem.  I have a 750GB drive in my secondary server at home, no
special hardware, no crazy hoops to jump through:
=============================



=============================
From:    Tom Miller <[log in to unmask]>
> I have read some statements
> that Linux does not use the bios and is therefore immune to any such
> problems, but I find that hard to believe.
It may be hard to believe, but it's true. Once the BIOS has handed over

operation of the computer to the kernel, Linux doesn't care what size 
BIOS thinks the hard drives are. It communicates directly with the IDE

controller to get the information from the drive. It uses an IDE
command 
called, surprisingly, IDENTIFY DRIVE to get the CHS, LBA, or 48-bit LBA

depending on the size of the drive. So, you can plug pretty much any
ATA 
drive into your old computer and Linux will see the correct size. 
=============================

Ok, I'm starting to believe this :)   Paraphrasing, does the command
and response go through the "controller" in the machine through to the
real electronics that run the drive?   It almost has to be working, as
Linux partitioning tools reported the full size of the drive.  Given
that much, is safe to assume Linux can access it?



=============================
You 
will want a /boot partition that is in the area the BIOS can see, near

the beginning of the disk. I do agree with the previous comment here 
that if you expect high performance from the disks you want to put them

on separate controllers, not just separate channels on the same
controller.
=============================

That is an interesting point.  In one of my experiements, I put a 500GB
drive in my new arrival, and Ubuntu installed w/o incident, only to fail
on restart with a GRUB error.  What do I need to do to ensure /boot is
in the right place?  I believe I left that drive with RedHat 8, so it
should be simple enough to try again with Ubuntu.  The machine currently
has a 20GB hand-me-down drive that appears to be working just fine. 
Before that, it had an 80GB drive, of uncertain history, that started
reporting bad sectors shortly after returning to use; it came from a
computer that suspiciously stopped working, and that I had never used. 
One could argue that all this stuff starts happening when a drive is
installed in the machine I am trying to use; let me know if you want to
assert that, and better yet, if you have an idea for isolating a
problem, hopefully w/o destroying any good hardware :)

Thanks,

Bill




Wilhelm K. Schwab, Ph.D.
University of Florida
Department of Anesthesiology
PO Box 100254
Gainesville, FL 32610-0254

Email: [log in to unmask]
Tel: (352) 846-1285
FAX: (352) 392-7029