devfs debate

Andrew J. Anderson (andrew@db.erau.edu)
Wed, 5 Aug 1998 09:28:16 -0400 (EDT)


Terry L Ridder wrote:

> Do I like the way /dev/sd? is rearranged when I pull a hard drive out
> of the SCSI expansion box? Well no, but I have learned to live with
> it for the time being.

Let's see now, I'm running a production server, and I need to add a SCSI
drive:

Under the current naming convention I add the new drive, and then spend an
hour determining how the SCSI chain has re-ordered itself and where the
various /dev/sd? devices I need have relocated themselves to, and I clean
up fstab and reboot.

OR

I add a new drive and reference the new controller/ID based name and I'm
back up in 5 minutes, I add new fstab entries at my leisure.

I would prefer the latter, of course, since most user/customers don't like
to be without services.

> We need to keep the "KISS" principle in mind. While the naming scheme
> of dev_fs may be logical it is not simple.

These are the same screams of pain that SUN SunOS users use during the
transition to Solaris. The difference there is that Solaris *forces* you
to use the new naming scheme, devfs does not!

> /dev/sda, /dev/sda[1-15] is simple.

Simple and IMNSHO totally inadequate as we grow into the so called
"enterprise computing" arena. Tell me without digging through
/proc/scsi/scsi conslusively what /dev/sdc is? You can't unless you're
intimately familiar with the machine. Most of us running in a production
environment don't have that luxury.

Now, tell me what /dev/dsk/c0t0d0s0 is -- easy controller 0 target 0
device 0 slice 0. The only ambiguity is which controller is 0? That is
much easier to determine (using dmesg) than /dev/sdc (digging around in
/proc) for someone who works with multiple unices, not just Linux. Try
managing multiple servers with a half dozen scsi busses and you'll learn
to appreciate the SYSV deice scheme more.

You yourself claim that we need to support database vendors with SCSI and
filesystems. I have seen people mention 288 gigs of disk space using 18
gig drives. Ok, sure, go right ahead. Performance will suck massively.

If someone is serious about Oracle, we're talking _terabytes_ of disk
space, not mere gigabytes. We've got several hundred disks attached to
our SUN servers to do this. Hmmmm. OK, let's extend /dev/sd? like the
/dev/pty?? devices so now we have /dev/sd?? to handle that many drives.

Yeah, right! I hope that it's painfully obvious that /dev/sd? has to
change at some point, the only question is when, and providing a migration
path now is excellent timing.

> Richard, I have read your FAQ where the naming scheme for SCSI disks
> is described and it screams "ugly".

It may not be as "simple" as /dev/sda[1-15], but it also "screams"
industry (SYSV) standard -- you know Solaris, HP-UX, IRIX, SCO, etc, etc.

=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=
Andrew Anderson http://amelia.db.erau.edu/~andrew/
if(!(family_tree=fork())){redneck=TRUE;}

-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.rutgers.edu
Please read the FAQ at http://www.altern.org/andrebalsa/doc/lkml-faq.html