Subject: Re: Config ...
To: C Kane <ckane@best.com>
From: Colin Wood <cwood@ichips.intel.com>
List: tech-kern
Date: 08/21/1998 16:24:02
C Kane wrote:
> somebody wrote:
> >> >     More common: you turn of one of your disks/tapedrives. 
> >> >     Are you going to renumber the remaining ones?
> 
> Probably completely separate from the issue of PCMCIA devices:
> 
> Something I've long disliked about NetBSD is the numbering
> scheme used to name disks (and tapes and many other peripherals).
> 
> Why is the first disk called "sd0" and the second disk "sd1"
> regardless of SCSI ID or controller?  I think even a new naive 
> user would understand that a disk on SCSI ID 4 is called sd4.

but what do you call the disk with SCSI ID 4 on SCSI controller 2?

you can always wire down your SCSI ID numbers if you want, tho.
 
> Why not use some unique identifier "XXX" for the controller and call
> devices /dev/cXXXdID where ID is the actual SCSI ID?  Then if a
> drive goes bad or if devices are moved and the system reboots the
> fstab will still match.

i suppose the above might work, but the only way /etc/fstab is still going
to work is if you wire down your controllers.  what if i grab two
identical host adapters...how do i get unique indentifiers/device names
out of them?

> We have HP systems at work which have two SCSI chains:  FWD for
> the internal disks and SE for the external devices.  We attach and
> detach CD-ROMs and DAT drives and occasionally disks from the SE
> chain all the time.  Since HP assigns device names based on controller
> and SCSI ID there are never any problems about using the correct name
> for any particular device.  You don't even have to re-scan the SCSI
> bus so the OS finds new devices -- you simply use them and it works.
> 
> I don't think I'd ever expect that I could take my /u1 disk, for example,
> move it to a different SCSI ID and a different controller and expect
> that any UNIX would automatically find it and mount it as /u1 by itself,
> because I'm not sure that I'd ever really want that to happen anyway.
> But I would expect that if a SCSI disk died and I reboot my machine,
> that all the disks that are working mount in the correct place.  Now I 
> know the device names can be "nailed down" for all the SCSI peripherals,
> but I have three wide chains and that last drive would be called /dev/sd47
> and I'm not sure that NetBSD can deal with that.

yeah, that would be a pain....

later.

-- 
Colin Wood                                 cwood@ichips.intel.com
Component Design Engineer - PMD                 Intel Corporation
-----------------------------------------------------------------
I speak only on my own behalf, not for my employer.