SUMMARY: formatting 7200 rpm disks under 4.1.3_U1

From: Adam Epstein (ase+@CMU.EDU)
Date: Tue Sep 19 1995 - 16:56:18 CDT


Original query, summary, and summarized replies follow:

| I'm trying to format a Quantum XP34301S on a SPARC2 running 4.1.3_U1
| and /usr/etc/format tells me that the acceptable range for rpm is from
| 3000 to 6000. The XP34301S is a 7200 rpm disk. Can anybody tell me
| how to get around this? I don't see an explanation in any of the
| FAQs. Hopefully I didn't miss it.
|
| Also, I saw the following posted to the list by Paul Hostrup-Jessen
| (phj06@bk.dk):
|
| >Subject: SUMMARY: format.dat for Quantum XP34301S
| >
| >I was looking for the "format.dat" for the "Quantum Grand Prix XP34301S"
| >
| >Thanks to Jim Mintha (mintha@geog.ubc.ca) who used the "scsiinfo" program to
| >achieve the characteristics (available from ftp.cdf.toronto.edu) here it is:
| >
| >
| >disk_type = "QUANTUM XP34301" \\
| > : ctlr = SCSI : fmt_time = 9 \\
| > : trks_zone = 20 : atrks = 0 : asect = 10 \\
| > : ncyl = 3561 : acyl = 2 : pcyl = 4076 : nhead = 20 : nsect = 118 \\
| > : rpm = 7200 : bpt = 69384
| >
| >I already have my disk up and running!
|
| However, Quantum supplied me with the following numbers:
|
| ncyl = 4080 : acyl = 2 : pcyl = 4082 : nhead = 20 : nsect = 103
|
| Why the difference?

  The key answer was that the 3000rpm - 6000rpm limit is enforced only
if the parameters are entered interactively when running format.
(Thanx to John Valdes and Jochen Bern who got it right, although I had
already discovered the answer on my own).

  Also useful was the information that rpm (and some of the other
parameters) are either non-critical or no longer applicable.

  And finally, after successfully formating the drive, I tried
building filesystems - first on one 4.3G partition that covered the
whole disk (failed w/seek error), next on two 2.15G partitions (failed
w/seek error), and finally on three 1.4G partitions (succeeded).
Hmmm. 2.15G is just over 2.147G which is where 32 bit unsigned ints
top out. Hmmm...

  Thanx to all who responded.

  -Adam
_____________________________________________________________________________
From: Chris Royle <Chris.Royle@elmail.co.uk>

The speed of the drive is only used by SunOS for optimization of access.
We regularly ship 7200rpm 4GB barracudas formatted supposedly "at 3600
rpm". It makes no difference really. SunOS will just lay things out on
the disc marginally differently, that's all.
_____________________________________________________________________________
From: freyr!pakers@netcom.com (Phillip Akers)

Don't worry about the RPM... I set mine to 5400 and it worked just
fine... I do not believe it is even necessary any more and was only
used with the older controllers... also the numbers are not super
critical unless you exceed the capacity of the drive...
_____________________________________________________________________________
From: Chin Fang <fangchin@jessica.Stanford.EDU>

I may not be able to offer you a solution as I don't have experience
with Grand Prix yet. However, I can see why Quantum gives you different
data than the posted ones.

[some removed]

Per Quantum's formula for calculating pcyl:

pcyl == MLBA/(nhead * nsec). I don't know what MLBA means, I assume it's
the same as logical blocks that the drive has. Taking what I have from
Quantum data sheet, for your drive, nhead = 20, nsect = 118, LB = 8410200

Using the above formula, npcyl = 8410200 / ( 20 x 118 ) ~ 3564.

Quantum also uses the relation ncly = pcyl - 2 ==> ncly = 3562
_____________________________________________________________________________
From: Glenn.Satchell@uniq.com.au (Glenn Satchell - Uniq Professional Services)

Just use 6000 rpm, it doesn't really make any difference.
_____________________________________________________________________________
From: John Valdes <valdes@geosun.uchicago.edu>

It's not possible to specify an rpm > 6000 when entering drive specs
interactively with format. Instead, create a format.dat entry for the
drive and then run format; it will gladly except an rpm 7200 this way.

Having to specify all these parameters to format is pretty archaic;
for SCSI disks, many of these parameters (eg, nsect) aren't even
constant across the entire surface of the disk. The primary purpose
for SCSI disks is to have ncyl*nhead*nsect be as close as possible to
the total number of blocks on the disk. Both format.dat entries
specify similar number of blocks, so either should be fine.
_____________________________________________________________________________
From: bern@penthesilea.uni-trier.de (Jochen Bern)

Try entering it into format.dat right away. 'format' sometimes accepts Data
from format.dat that it wouldn't accept from manual Input (e.g., acyl=1).

With SCSI Disks, all Communication between Computer and Disk uses
absolute Sector Numbers, nothing else. It is wise to keep the Figures
*close* to "real Values", but the one and only Thing that count at last
is whether you add up to a Number of Sectors that the Disk can offer.
The first Set of Parameters makes 8,403,960 Sectors available to Users,
the second one 8,404,800 - slight Gain.
_____________________________________________________________________________
From: roland@netcom.com (Paul Roland)

This is our working entry:

disk_type = "Quantum XP34301" \
        : ctlr = SCSI : fmt_time = 4 \
        : ncyl = 3561 : acyl = 2 : pcyl = 3563 \
        : nhead = 20 : nsect = 118 \
        : rpm = 7200 : bpt = 60416

It, too, is supplied by Quantum, also. I can't explain the discrepency.
Format doesn't complain with our installation of 4.1.3_U1. I've never seen
a totally consistent format.dat entry for any modern drive from any two
sources.

As for the 7200rpm value, I believe it only refers to estimates about
format time.



This archive was generated by hypermail 2.1.2 : Fri Sep 28 2001 - 23:10:34 CDT