Subject: Re: partition bookkeeping
To: Ignatios Souvatzis <email@example.com>
From: Oleg Polyanski <firstname.lastname@example.org>
Date: 09/23/1999 00:34:57
>>>>> "IS" == Ignatios Souvatzis writes:
>> >> I'm not sure that it makes our life easier. Right now we have
>> floating >> disk naming scheme - disk detected first will get number `0'
>> even if it >> has, for example, SCSI ID 5 so when I add another disk with
>> SCSI ID >> number 2 (for example, I would like to add large /opt file
>> system) it >> becomes first (i.e. sd0), not the former `first' disk.
>> You cannot mount >> root fs without editing your `/etc/fstab'. It's
>> weird. Disk naming >> scheme must depend on device physical properties
>> (SCSI ID, master/slave >> or something like that).
IS> This won't help you to mount root. You have to _know_ at _runtime_ which
IS> one it is. As this isn't generally possible (at best, the kernel knows
IS> it), install scripts tend to mount kernfs and then mount /kern/rootdev.
>> Sorry, forgot to say `mount root fs for read/write'.
IS> Yes, this was evident, and is exactly what I'm referring to.
IS> Maybe I should add that for normal booting, root devices tend to not
IS> change their name all the time, so having the sysadmin edit fstab once
IS> at installation or reconfiguration time is good enough.
Consider my disk has name `sd0' (because it was detected first during
the kernel booting). Later I will add ZIP drive and IT will get the
name `sd0' and hard disk will become `sd1'. Bad surprise, manual
intervention required. Something like `/dev/scsi/dsk/c0d0p0' as root
fs unique identifier will help.