From owner-freebsd-questions Sat Aug 15 15:14:48 1998 Return-Path: Received: (from majordom@localhost) by hub.freebsd.org (8.8.8/8.8.8) id PAA21091 for freebsd-questions-outgoing; Sat, 15 Aug 1998 15:14:48 -0700 (PDT) (envelope-from owner-freebsd-questions@FreeBSD.ORG) Received: from phoenix.welearn.com.au (suebla.lnk.telstra.net [139.130.44.81]) by hub.freebsd.org (8.8.8/8.8.8) with ESMTP id PAA21086 for ; Sat, 15 Aug 1998 15:14:44 -0700 (PDT) (envelope-from sue@phoenix.welearn.com.au) Received: (from sue@localhost) by phoenix.welearn.com.au (8.8.5/8.8.5) id IAA04939; Sun, 16 Aug 1998 08:14:00 +1000 (EST) Message-ID: <19980816081357.54810@welearn.com.au> Date: Sun, 16 Aug 1998 08:13:57 +1000 From: Sue Blake To: djv@bedford.net Cc: freebsd-questions@FreeBSD.ORG Subject: Re: disk confusion References: <19980816015957.37974@welearn.com.au> <199808151724.NAA12486@lucy.bedford.net> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii X-Mailer: Mutt 0.88e In-Reply-To: <199808151724.NAA12486@lucy.bedford.net>; from CyberPeasant on Sat, Aug 15, 1998 at 01:24:08PM -0400 Sender: owner-freebsd-questions@FreeBSD.ORG Precedence: bulk X-Loop: FreeBSD.ORG On Sat, Aug 15, 1998 at 01:24:08PM -0400, CyberPeasant wrote: > Sue Blake wrote: > > To do the installation (from CD) I disabled the IDE (wdc0) and > > unplugged sd0, making the working disk take the name sd0 temporarily. > > Now of course it's called sd1 again. > > Ouch. > > I bet "real" sd1's /etc/fstab is just chock full of references to > sd0sX Argh, of course, thanks, it was a "memory" problem after all :-) > > Yes it booted fine as sd0 before I plugged the first SCSI disk back in. > > Right, sd0... Repeat last comment, s/bet/really bet/ > > At this point, I would boot FreeBSD from any but the sd1 system; > I would fsck the sd1s1X partitions. > > Then I would mount the sd1s1a on /mnt, and poke around in > /mnt/etc, looking for sd0 [fstab :) ] where it should be sd1. Did that, but the gremlins are one step ahead of me. It still won't let me in coz there's no /dev/sd1s1a, only sd1 (but of course they're all present for sd0!). It starts complaining about not finding swap and then gets fussy about everything else. I tried changing to /dev/sd1a in fstab to fool it long enough to make the devices, but it still wouldn't let me mount /. I don't suppose I can make the right devices on sd1 by mounting the drive after booting from sd0, can I? (If so you'd better give me the MAKEDEV incantation too so I can't screw that up) > I just went through this drill yesterday, more or less. (I was > using a scsi controller trick to accomplish the same as unplugging > the drive.) > > Alas, this stuff arises because of convenience and user friendliness. > > You have to jump through hoops to get the ancient behavior (the > one the gods intended men to use), where SCSI ID 4 becomes sd4, > regardless of what other devices are on the SCSI bus. My two drives > are ID 0 and ID 2, but show up as sd0 and sd1. If I tell the > controller to boot from ID 2, then ID 2 becomes sd0 and ID 0 becomes > sd1. This is, of course, user friendly. Grrrr. It's real friendly, > until you want to boot the rescue on ID 2. This will be sd0 if > old ID 0 has failed due to hardware (can't be detected by the controller) > or sd1 if ID 0 is still alive at the SCSI level. Yeah, works predictably until something out of the ordinary happens, like hardware failure. I have details of all my SCSI devices tabulated on a card, all the names and numbers they are allocated by the various gods, where they are on the chain and what identifying marks are visible without removing them. This card is inside the case. If I ever lost it, life wouldn't be worth living :-) > The hoops to jump through involve, IIRC, "wiring" the ID to a > specific device by configuring yet another new kernel. (Your problem > won't require this, I think). Thankfully no. It's a bit hard to rebuild a kernel on a system that won't boot until the kernel's rebuilt. -- Regards, -*Sue*- To Unsubscribe: send mail to majordomo@FreeBSD.org with "unsubscribe freebsd-questions" in the body of the message