Very Computer

: Hi there,

:       can anybody help me in the following subject:
: I'm trying to setup an high-end server (i386) running Linux. What I need
: is to use 30 Microplois 9.1GB disks connected to two AHA2940 UW
: controllers. Since each disks requires 16 inodes/minor numbers in /dev,
: I am restricted to 16 disks sda...sdp. Is there any possibility to
: reduce the number of partitions per disk (I only need one!) in order
: do have some new minors for the extended disks? Or can I get a new
: major number for the second SCSI controller? Does anybody has experience
: in this topic?

: Hope somebody can help me, best regards...

There's a related story in the newest LJ.
I suggest creating new device files with 'mknod', just continue the array of
minor numbers.

--
Robert Bergfors

---
When a cat is dropped, it always lands on its feet, and when toast is dropped, it
always lands with the buttered side facing down. I propose to strap buttered
toast to the back of a cat; the
two will hover, spinning inches above the ground. With a giant buttered cat
array, a high-speed monorail could easily link New York with Chicago.
        - OMNI Magazine, "Best of Humour"-winner

-z

--

# UW-Madison ECE GradStd.  CAE UnixStaff.  DigiGeek.  CB-1. "Who is John Galt?"

--
To reply via mail, please remove the obvious from the email address.

Quote:>another solution would be to adopt a new naming convention for devices.
>personally I like the version that solaris uses:  c#t#d#s#  for controller,
>target (scsi id), device (lun), and partition.  It's simple, tells you what
>is doing what, and you can avoid the 'added a new disk so my old sdb is now
>sdc' problems ...

        -Kenny

--
Kenneth R. Crudup, Unix & OS/2 Software Consultant, Scott County Consulting

16 Plainfield St, Boston, MA 02130-3633   +1 617 983 9410      Fax
        "Linux, between Playstation and Craystation" -attribution unknown

A brief look through sd.[ch] makes me think that getting an extra
bit for the target number (and reducing the max number of partitions
to 8) might be as simple as changing DEVICE_NR() in blk.h and
changing a bunch of '4's to '3's throughout sd.c. Better yet,
make it a #define ... sd.c has too many magic numbers for my taste.
The usual disclaimers apply, of course. :-)

--

 PGP 0x27F772C1: 0C 0D 10 D5 FC 73 D1 35  26 46 42 9E DC 6E 0A 88

Quote:> Except that the problem isn't the lack of named for /dev/ files, it's
> the lack of device minor numbers. A minor number is 8 bits; if 4
> bits are used for the partition number (that is, 16 partitions possible

So ...  I agree with you on the major/minor changes.  I just wanted to throw
in the comment that a naming convention change would be useful as well.

--
To reply via mail, please remove the obvious from the email address.

... and who will ever need more than 1024 cylinders on a hard disk?

I can see what you are saying is a fix for this persons problem, but in
the
future I bet other people will strike a limit on 32 disks with 8
partitions.
Maybe it's time to allocate a second major number for SCSI disks, and
the code
be written so that a third, fourth, etc. can be added pretty easily in
the future.

Just my $0.02 worth.

Brian

Quote:

> So ...  I agree with you on the major/minor changes.  I just wanted to throw
> in the comment that a naming convention change would be useful as well.

> --
> To reply via mail, please remove the obvious from the email address.

Quote:> Except that the problem isn't the lack of named for /dev/ files, it's
> the lack of device minor numbers. A minor number is 8 bits;

Neil