From owner-freebsd-hackers Fri Nov 28 19:40:04 1997 Return-Path: Received: (from root@localhost) by hub.freebsd.org (8.8.7/8.8.7) id TAA29638 for hackers-outgoing; Fri, 28 Nov 1997 19:40:04 -0800 (PST) (envelope-from owner-freebsd-hackers) Received: from trojanhorse.ml.org (mdean.vip.best.com [206.86.94.101]) by hub.freebsd.org (8.8.7/8.8.7) with ESMTP id TAA29617 for ; Fri, 28 Nov 1997 19:40:01 -0800 (PST) (envelope-from jamil@trojanhorse.ml.org) Received: from localhost (jamil@localhost) by trojanhorse.ml.org (8.8.8/8.8.5) with SMTP id TAA03958; Fri, 28 Nov 1997 19:32:31 -0800 (PST) Date: Fri, 28 Nov 1997 19:31:11 -0800 (PST) From: "Jamil J. Weatherbee" To: Satoshi Asami cc: hackers@freebsd.org Subject: Re: Drive Mirroring In-Reply-To: <199711290256.SAA22913@silvia.HIP.Berkeley.EDU> Message-ID: MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: owner-freebsd-hackers@freebsd.org X-Loop: FreeBSD.org Precedence: bulk Would a software RAID implementation be able to cope with having a drive removed? When the drive is reconnected it should simply take care of bringing it up to date or completely remaking it from the current master. There is no reason that this should not be a standard feature, as SCSI disks die all the time. I want to have a solution where the probability of total failure is reduced exponentially. Namely you can pull off and insert drives as needed without having to think about it, a total no-brainer for the kind of people who might be using these systems. Now I looked at some adaptec SCSI adapters that are supposed to do this in hardware, but I didn't really feel like buying a $900 scsi adapter to see if it properly functions with freebsd. Checksums should be kept on CCD blocks to insure that the contents of a corrupted drive are not being mirrored over. I keep dat tapes like everyone else, but have you ever tried restoring a completely wiped system off of a dat tape? Especially when the drive is remote over an ethernet? You end up having to install enough of a system so that you can do the recovery. I want to stack 3 triple redundant SCSI drives and be able to smash two of them with a sledge hammer without a system glitch (I'll watch out for pass through cables). On Fri, 28 Nov 1997, Satoshi Asami wrote: > * One problem with this: If the system is rebooted with only one drive it > * crashes, second you cannot do a out of the box install on ccded drives so > * they are pretty useless tom me for anything but data. > > I can't help you about the second one, but you can make it boot > (manually) by editing /etc/ccd.conf (remove the mirror flag and the > dead partition). > > Satoshi >