From mboxrd@z Thu Jan 1 00:00:00 1970 Subject: Re: [linux-lvm] Urgent, Please help LVM-krash From: Erik =?ISO-8859-1?Q?B=E5gfors?= In-Reply-To: <20011217122223.A28720@sistina.com> References: <1008250625.4350.5.camel@detrius> <20011217122223.A28720@sistina.com> Content-Transfer-Encoding: quoted-printable Message-Id: <1008589296.18947.4.camel@detrius> Mime-Version: 1.0 Sender: linux-lvm-admin@sistina.com Errors-To: linux-lvm-admin@sistina.com Reply-To: linux-lvm@sistina.com List-Help: List-Post: List-Subscribe: , List-Unsubscribe: , List-Archive: Date: Mon Dec 17 05:40:01 2001 List-Id: Content-Type: text/plain; charset="iso-8859-1" To: Linux LVM Mailinglist m=E5n 2001-12-17 klockan 12.22 skrev Heinz J . Mauelshagen: > On Thu, Dec 13, 2001 at 02:37:04PM +0100, Erik B=E5gfors wrote: > > Hi all! > >=20 > > Yesterday I krashed my lvm. This is the first time I'm having real > > problems with it. > >=20 > > This is what happened. > >=20 > > I was going to add a new lv and used the lvcreate-command as usual.=20 > > This time I forgot the -n switch and when I realized that I hit ctrl-c.= . > > BAD idea! >=20 > Erik, >=20 > that's actually starneg, because the tools intercept ctrl-c and the like = :-( Hmm.. well .. it did break the program for me.. Maybe it's an old version with mandrake 8.1?? I know that the kernel lvm-stuff is NOT the latest version. Is it ok to run an old version or do I risk damage to the system?? =20 > > Now my vg is gone. I can't get it back whatever I try. Any help would > > be VERY appreciated. > >=20 > > A vgscan just sais "no volume groups found" or something like that.=20 > > doing a pvdisplay on the pv's in my vg gives me that they should be in > > "diskvg" (which is correct). I've tried doing vgcfgrestore from the > > backups in /etc/lvmconf but that doesn't help. vgscan still doesn't > > find it. >=20 > Did you run "pvcreate -ff" on all of your PVs belonging to diskvg *before= * > you run vgcfgrestore on all of them? vgscan afetrwards. At first I didn't, but I tried that later. Same thing. =20 > >=20 > > I looked at the HOWTO and found a program named uuid_fixer that looked > > like it could solve the problem but it died with an "illegal > > instruction" regardless of if I used the precompiled version or compile= d > > it myself. > >=20 > > This is on a mandrake 8.1-box with lvm, xfs, reiserfs and devfs so ther= e > > are alot of things to think about. > >=20 > > When doing stuff I did the following first: > > reboot into single user, remount / rw, mount /proc, restore some > > standar-stuff into /dev, and start devfsd > >=20 > > After that I tried vgcfgrestore, vgscan, uuid_fixer and just about > > everything I could think or. >=20 > I think it was just the lack of pvcreate before vgcfgrestore. Perhaps. But I think there was another problem. I've had a few small problems with xfs under mandrake 8.1. Sometimes I've lost some files. I really don't know why but the data in them have been corrupted or they were 0 bytes in size. I have been really mean to the computer and killing the power on it while it was running and so on. But that was never a problem under reiserfs so I'm back with 100% reiserfs now. I think the files in /etc/lvmconf were strange. =20 > >=20 > > Please please please help me. Can anyone solve this. There are over 6= 0 > > GB of stuff on this computer and it needs to work TODAY :( > > Like I said I needed that computer back the same day. I ended up reinstalling the box. Loosing the data I had on it. Luckily most data was either not important or could be restored from backup. The only thing I really lost was the system with all it's services and the like.=20 So I guess I need to resetup samba/appleshare/web/nfs/dns and so on. =20 /Erik --=20 Erik B=E5gfors | erik@bagfors.nu Supporter of free software | GSM +46 733 279 273 fingerprint: 6666 A85B 95D3 D26B 296B 6C60 4F32 2C0B 693D 6E32