Date: Mon, 6 Mar 2006 15:19:47 +0100 From: "Olivier Cochard" <olivier@freenas.org> To: "Lukas Ertl" <le@freebsd.org> Cc: freebsd-geom@freebsd.org Subject: Re: Gvinum RAID5 volume don't work after a simple reboot Message-ID: <3131aa530603060619o6fd23c8bl@mail.gmail.com> In-Reply-To: <Pine.WNT.4.64.0602272234530.1080@korben> References: <3131aa530602271250r458dd9fep@mail.gmail.com> <Pine.WNT.4.64.0602272206350.1080@korben> <3131aa530602271316n479af14fi@mail.gmail.com> <Pine.WNT.4.64.0602272227470.1080@korben> <3131aa530602271330t5b94992ei@mail.gmail.com> <Pine.WNT.4.64.0602272234530.1080@korben>
next in thread | previous in thread | raw e-mail | index | archive | help
I can reproduce the problem (tested under VMware). It's a long mail, But i can reproduce at each time! For resume: When I create a RAID 5 volume, I must run newfs twice for have a permanent raid volume (if not, I loose the RAID 5 when I reboot the PC). Details: - VMware workstation 5.5.1 - Tested with I'm using a small FreeBSD system (6.0 with FreeNAS 0.6 and 6.1beta with FreeNAS 0.63) - Have 3 disks of 100Mb each (ad0, ad1, ad3) --------------------------------------------------- Step 1: Initialized and format the 3 disks with theses commands: /sbin/fdisk -I -b /boot/mbr ad0 /bin/dd if=3D/dev/zero of=3D/dev/ad0s1 bs=3D32k count=3D16 /sbin/bsdlabel -w ad0s1 auto Editing the bsdlabel for each: - replacing c: by a: - changing "unused" to "vinum" --------------------------------------------------- Step 2: Create this raid.conf file: drive disk_ad0 device /dev/ad0s1 drive disk_ad1 device /dev/ad1s1 drive disk_ad3 device /dev/ad3s1 volume raid5 plex org raid5 256k sd length 102M drive disk_ad0 sd length 102M drive disk_ad1 sd length 102M drive disk_ad3 --------------------------------------------------- Step 3: Create the volume: /sbin/gvinum create raid.conf --------------------------------------------------- Step 4:Start the volume /sbin/gvinum start raid5 For information at this state, all it's ok: 3 drives: D disk_ad3 State: up /dev/ad3s1 A: 0/102 MB (0%) D disk_ad1 State: up /dev/ad1s1 A: 0/102 MB (0%) D disk_ad0 State: up /dev/ad0s1 A: 0/102 MB (0%) 1 volume: V raid5 State: up Plexes: 1 Size: 204 MB 1 plex: P raid5.p0 R5 State: up Subdisks: 3 Size: 204 MB 3 subdisks: S raid5.p0.s2 State: up D: disk_ad3 Size: 102 MB S raid5.p0.s1 State: up D: disk_ad1 Size: 102 MB S raid5.p0.s0 State: up D: disk_ad0 Size: 102 MB --------------------------------------------------- Step 5: Create filesystem on this volume /sbin/newfs -U /dev/gvinum/raid5 --------------------------------------------------- Step 6: I can mount it wihtout problem, and copy lot's of file on it withou= t problem: $ mount /dev/md0 on / (ufs, local) devfs on /dev (devfs, local) /dev/fd0 on /cf (msdosfs, local, read-only) /dev/gvinum/raid5 on /mnt/raid5mnt (ufs, local, soft-updates) touch /mnt/raid5mnt test.txt --------------------------------------------------- Step 7: I reboot the system. The rebuilding of the raid volume take a long time... (1 minutes for 3 disk of 100Mb each). --------------------------------------------------- Step 8: Impossible to mount the raid5 volume: $ mount /dev/gvinum/raid5 /mnt/raid5 mount: /dev/gvinum/raid5 on /mnt/raid5: incorrect super block $ fsck -t ufs /dev/gvinum/raid5 Cannot find file system superblock fsck_ufs: /dev/gvinum/raid5: can't read disk label ** /dev/gvinum/raid5 ioctl (GCINFO): Inappropriate ioctl for device Loosing all the files.... !!!! --------------------------------------------------- Step 9: Re-create the filesystem /sbin/newfs -U /dev/gvinum/raid5 --------------------------------------------------- Step 10: Then I can re-mount a filesystem without problem (but without the file!) $ mount /dev/md0 on / (ufs, local) devfs on /dev (devfs, local) /dev/fd0 on /cf (msdosfs, local, read-only) /dev/gvinum/raid5 on /mnt/raid5mnt (ufs, local, soft-updates) And create file.. touch /mnt/raid5mnt titi.txt --------------------------------------------------- Step 11: Reboot for the second time (after re-create the filesystem a secon= d time) No problem! I can mount the RAID volume without problem. Regards, Olivier -- Olivier Cochard FreeNAS main developer http://www.freenas.org Skype: callto://ocochard 2006/2/27, Lukas Ertl <le@freebsd.org >: > > On Mon, 27 Feb 2006, Olivier Cochard wrote: > > > Yes, the volume is UP but I can't mount it, and the fsck report the > error > > message previously post. > > Hm, actually, no idea. Maybe the first superblock got lost somehow, and > you can find an alternate superblock to mount the FS. Using "newfs -N" > should print out how a FS would be created without doing it, so you can > find the locations of the alternate superblocks. > > regards, > le > > -- > Lukas Ertl http://homepage.univie.ac.at/l.ertl/ > le@FreeBSD.org http://people.freebsd.org/~le/<http://= people.freebsd.org/%7Ele/> >
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?3131aa530603060619o6fd23c8bl>