Re: Broken softraid0

On Wed, 2022-09-07 at 18:18 +0100, Zé Loff wrote:
>
> On Wed, Sep 07, 2022 at 06:14:59PM +0100, Zé Loff wrote:
> > On Wed, Sep 07, 2022 at 06:42:54PM +0200, Danny König wrote:
> > > Hi,
> > >  
> > > first time I need your support regarding a broken softraid0.
> > > After a
> > > normal reboot the softraid0 (RAID 1, two disks sd0 and sd1,
> > > softraid0
> > > as sd5) was broken what means no mounts, no nothing.
> > >
> > >  
> > > root@core-fs:~ => disklabel -p g /dev/sd0c
> > > # /dev/sd0c:
> > > type: SCSI
> > > disk: SCSI disk
> > > label: ST16000NM003G-2K
> > > duid: b1c256d6dc122fa7
> > > flags:
> > > bytes/sector: 512
> > > sectors/track: 63
> > > tracks/cylinder: 255
> > > sectors/cylinder: 16065
> > > cylinders: 1945332
> > > total sectors: 31251759104 # total bytes: 14902.0G
> > > boundstart: 0
> > > boundend: 31251759104
> > > drivedata: 0
> > > 16 partitions:
> > > # size offset fstype [fsize bsize cpg]
> > > a: 14902.0G 0 RAID
> > > c: 14902.0G 0 unused
> > >
> > >
> > >
> > > root@core-fs:~ => disklabel -p g /dev/sd1c
> > > # /dev/sd1c:
> > > type: SCSI
> > > disk: SCSI disk
> > > label: ST16000NM003G-2K
> > > duid: acedcb2d1684fb18
> > > flags:
> > > bytes/sector: 512
> > > sectors/track: 63
> > > tracks/cylinder: 255
> > > sectors/cylinder: 16065
> > > cylinders: 1945332
> > > total sectors: 31251759104 # total bytes: 14902.0G
> > > boundstart: 0
> > > boundend: 31251759104
> > > drivedata: 0
> > > 16 partitions:
> > > # size offset fstype [fsize bsize cpg]
> > > a: 14902.0G 0 RAID
> > > c: 14902.0G 0 unused
> > >
> > >
> > >
> > > root@core-fs:~ => disklabel -p g /dev/sd5c
> > > # /dev/sd5c:
> > > type: SCSI
> > > disk: SCSI disk
> > > label: SR RAID 1
> > > duid: 0000000000000000
> > > flags:
> > > bytes/sector: 512
> > > sectors/track: 255
> > > tracks/cylinder: 511
> > > sectors/cylinder: 130305
> > > cylinders: 239835
> > > total sectors: 31251758576 # total bytes: 14902.0G
> > > boundstart: 0
> > > boundend: 31251758576
> > > drivedata: 0
> > >
> > > 16 partitions:
> > > #                size           offset  fstype [fsize bsize  
> > > cpg]
> > >   c:         14902.0G                0  unused
> > >
> > >
> > >
> > >  
> > > root@core-fs:~ => bioctl softraid0
> > > Volume Status Size Device
> > > softraid0 0 Online 16000900390912 sd5 RAID1
> > > 0 Online 16000900390912 0:0.0 noencl <sd0a>
> > > 1 Online 16000900390912 0:1.0 noencl <sd1a>
> > >
> > >
> > >  
> > > Thats what we got, but no mounts are possible.
> > >  
> > > Any ideas how to become the data back available?
> > >  
> > > Thank you very much.
> > >  
> > > Regards,
> > > Danny
> > >
> >
> > sd5's disklabel looks pretty empty (duid included), but you already
> > know
> > that...
> > Do you have a backup disklabel, e.g., in /var/backups?


Yes, that's the output:

# /dev/rsd5c:
type: SCSI
disk: SCSI disk
label: SR RAID 1C
duid: 1c318a61d9aa3255
flags:
bytes/sector: 512
sectors/track: 63
tracks/cylinder: 255
sectors/cylinder: 16065
cylinders: 1945331
total sectors: 31251758576
boundstart: 0
boundend: 31251758576
drivedata: 0

16 partitions:
# size offset fstype [fsize bsize cpg]
c: 31251758576 0 4.2BSD 2048 16384 12960


>
> Actually, I'll just quote scan_ffs's manpage:
>
>      1.    Panic.  You usually do so anyways, so you might as well
> get it over
>            with.  Just don't do anything stupid.  Panic away from
> your
>            machine.  Then relax, and see if the steps below won't
> help you
>            out.
>
>      2.    Try to find your old disklabel by any other means
> possible.  This
>            includes printouts, backups (look in /var/backups/),
> screendumps,
>            and whatever other method you can think of.  The more
> information
>            you have, the better your chances are in recovering the
> disklabel
>            of the disk.
>
>      3.    Create a disklabel on the affected disk, which covers the
> whole
>            disk, and has at least one partition which covers the
> whole disk.
>            As the "c" partition usually covers the whole disk
> anyways, this
>            sounds like a good place to start.
>
>      4.    Run scan_ffs over this partition.  If you have any
> information
>            about the disklabel which used to exist on the disk, keep
> that in
>            mind while scan_ffs spews out its things.
>
>      5.    Use disklabel(8) to reconstruct the disklabel on the
> affected disk,
>            using all the information you gathered from scan_ffs and
> other
>            sources.
>
> (Make sure to read the rest of it, though)

I will read it. Thanks!