From owner-freebsd-questions Tue Sep 21 3:26:37 1999 Delivered-To: freebsd-questions@freebsd.org Received: from pluto1.daimler-benz.com (pluto1.daimler-benz.com [53.122.2.31]) by hub.freebsd.org (Postfix) with SMTP id 32C271509A for ; Tue, 21 Sep 1999 03:26:33 -0700 (PDT) (envelope-from norbert.meissner@daimlerchrysler.com) Received: by pluto1.daimler-benz.com; id MAA24215; Tue, 21 Sep 1999 12:24:36 +0200 Received: from unknown(53.113.75.10) by pluto1.daimler-benz.com via smap (V4.2) id xma024209; Tue, 21 Sep 99 12:24:32 +0200 Received: from c007fb66 (C007FB66.ut.str.daimler-benz.com [53.114.91.4]) by syssun11.rrz.debis.str.daimler-benz.com (8.8.8/8.8.8) with SMTP id MAA26400; Tue, 21 Sep 1999 12:25:27 +0200 (MET DST) Received: by localhost with Microsoft MAPI; Tue, 21 Sep 1999 12:25:59 +0200 Message-ID: From: Norbert Meissner To: "'Greg Lehey'" , "Meissner, Norbert" Cc: "'freebsd-questions@freebsd.org'" Subject: AW: vinum: strange behaviour and other things Date: Tue, 21 Sep 1999 12:25:57 +0200 Organization: DaimlerChrysler X-Mailer: Microsoft Internet E-Mail/MAPI - 8.0.0.4211 MIME-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Content-Transfer-Encoding: 7bit Sender: owner-freebsd-questions@FreeBSD.ORG Precedence: bulk X-Loop: FreeBSD.ORG Please limit your lines to about 75 characters. It makes them much easier to read, and avoids people with broken mailers making it completely illegible. Sorry, i have to work with M$ Outlook, no other chance for me here at DC. > 1. When I create a mirror volume a lot of error messages appeared > about that one plex is faulty. How many? I will send you the responses tomorrow, the machine is at home and my Internet connection is at work. It was about 4 lines or so. > Init the plexes doesn't help, but when I "start" the plexes they > will go in the "up" state. Is it possible to include this command in > the "create"-process? Yes. To quote the man page: <....> Oops, I've meant, when someone types "create -v ./mirror.cfg" the volume should be up without error messages and without need for typing additional commands. > 2. I was not able to make a dump from a vinum volume (raid5, mirror > and stripe tested). In the worst case (raid5) the whole system > freezes! Only the reset button shows effect. In the other cases dump > aborts with the message "bad magic number in sblock". You've sent in a PR with only part of this information. That doesn't help *one* *little* *bit*. Please specify *all* information when you enter a PR. Oops again, I've thought if a command doesn't work without affecting the system, it's a case for questions, otherwise when the whole system is pulled down it's a case for bugs. > I think that this is a problem with dump and the way it acesses the <...> Yes, you're correct, dump does some strange things. But it shouldn't cause vinum to do this. Do you think, that this problem will be solved in near future, aka, will someone spend some time to "patch" dump or will you change something on vinum? > 3. I'm missing documentation how to integrate a new disk in a vinum > volume for the case that one disk has failures. I built a mirrored > volume, then booted dos and deleted the partition info from one of > the disks. My data was after this accessible, I could write new data > on the remaining disk but I wasn't able to 'reintegrate' the disk > with no partition info in the mirrored volume. I tried to do this > with reconfiguring the partition and disklabel info. The next thing I > saw is that the drive was gone out of the vinum configuration ( the > line shows: "drive d1 device"). A create solved this. The volume was > up, but the plex was flaky, a "start mirror.p0" showed no effect, > "start mirror.p0.s0" states that it is busy. The same procedure on a > raid5 volume shows that the volume is degraded, the data is still > accessible but I didn' manage it to reintegrate the disk. At this > point I see for the case of a disk failure the only solution to back > up all data with tar, change the broken hd and restore. Or, maybe, > I'm wrong.... I'm still working on this one. At the moment, you need to recreate the objects manually. I'm not sure, what this means (yes, my knowledge of the english language is horrible!) Is there a way to bring in a brand new hd in an existing raid5 or mirror volume (flaky or degraded) as a replacement for a broken one? Or do you mean the solution I pointed out above? Norbert PS: Is the format of this mail right? +--------------------------------------------+ | _ _ __ __ _ | | | \| | | \/ |___(_)_______ _ ___ _ _ | | | .` |_ | |\/| / -_) (_-<_-< ' \/ -_) '_| | | |_|\_(_) |_| |_\___|_/__/__/_||_\___|_| | +--------------------------------------------+ Mercedes-Benz Stra?e 137 HPC G322 70546 Stuttgart Tel: +49-711-1754664 Fax: +49-711-178054664 norbert.meissner@daimlerchrysler.com To Unsubscribe: send mail to majordomo@FreeBSD.org with "unsubscribe freebsd-questions" in the body of the message