[linux-lvm] I/O error: dev 08:11, sector 0

[linux-lvm] I/O error: dev 08:11, sector 0

[Allan M. Wind]

[-- Attachment #1: Type: text/plain, Size: 1968 bytes --]

Hi,

Installed linux 2.4.6 + lvm 0.9.1beta7 with the kernel piece build
as a module (patched lvm and the kernel, got the user tools from
the .deb package of the right version).

# pvcreate /dev/sdb
# vgcreate -s16 vg1 /dev/scsi/host0/bus0/target9/lun0/disc
  (apparently, devfsd confuses lvcreate)
# lvcreate -L5g -nusr vg1
# mkfs -m0 /dev/vg1/usr
# lvcreate -L5g -nvar vg1
# mkfs -m0 /dev/vg1/var
# lvcreate -L24.03g -nhome vg1
# mkfs -m0 /dev/vg1/home

Updated /etc/fstab and:

# mount -a

Migrate data over, rebooted the box which does a:

# vgchange -a n

grub failed on me as the staging files were on lvm lv's
(that is probably a bug as one should be able to boot without
/usr), so booted off floppy.

Found that I had overlooked the check in /etc/init.d/lvm which
fails unless lvm was compiled in (yes, it was docmented but it
should probably be a tad smarter ;-) ).

# modprobe lvm-mod

(I might have had to do a vgscan & vgchange -a y)

# mount -a

so here is where I was looking for help... I get an I/O error
but things come up:

# vgck -v
vgck -- locking logical volume manager
vgck -- finding all volume group(s)
vgck -- checking volume group name "vg1"
vgck -- checking existence of volume group "vg1"
vgck -- reading volume group data for "vg1" from lvmtab
vgck -- checking volume group consistency  of "vg1" in lvmtab
vgck -- VGDA of "vg1" in lvmtab is consistent
vgck -- reading volume group data for "vg1" from physical volume(s)
 I/O error: dev 08:11, sector 0
vgck -- checking volume group consistency  of "vg1" on physical volume
vgck -- VGDA of "vg1" on physical volume is consistent
vgck -- unlocking logical volume manager

Is this a problem and how do I fix it?  This problem did not exist
before I rebooted.

Thanks in advance,


/Allan
-- 
Allan M. Wind			email: allanwind@mediaone.net
P.O. Box 2022			finger: awind@digit-safe.dyndns.org (GPG/PGP)
Woburn, MA 01888-0022
USA

[-- Attachment #2: Type: application/pgp-signature, Size: 232 bytes --]

Re: [linux-lvm] I/O error: dev 08:11, sector 0

[Heinz J. Mauelshagen]

Allan,

08:11 (/dev/sda11) shows that I/O error, you use /dev/sdb, which is 08:16
for your VG -> no problem for LVM.

Because /dev/sda11 seems to be configured, LVM reads it on its search for
physical volumes.

If you are not using /dev/sda11 for anything you are fine for the moment.

If there's a file system on /dev/sda11 *and* just the first sector is not
accessable, you can still happily access the filesystem, because it doesn't use
that sector at all.
But you should check and probably replace your sda drive to avoid
trouble in the future.

On Wed, Jul 18, 2001 at 11:17:29PM -0400, Allan M. Wind wrote:
> Hi,
> 
> Installed linux 2.4.6 + lvm 0.9.1beta7 with the kernel piece build
> as a module (patched lvm and the kernel, got the user tools from
> the .deb package of the right version).
> 
> # pvcreate /dev/sdb
> # vgcreate -s16 vg1 /dev/scsi/host0/bus0/target9/lun0/disc
>   (apparently, devfsd confuses lvcreate)

vgcreate? Yes, you need to use the full devfs paths for now.

Regards,
Heinz    -- The LVM Guy --

> # lvcreate -L5g -nusr vg1
> # mkfs -m0 /dev/vg1/usr
> # lvcreate -L5g -nvar vg1
> # mkfs -m0 /dev/vg1/var
> # lvcreate -L24.03g -nhome vg1
> # mkfs -m0 /dev/vg1/home
> 
> Updated /etc/fstab and:
> 
> # mount -a
> 
> Migrate data over, rebooted the box which does a:
> 
> # vgchange -a n
> 
> grub failed on me as the staging files were on lvm lv's
> (that is probably a bug as one should be able to boot without
> /usr), so booted off floppy.
> 
> Found that I had overlooked the check in /etc/init.d/lvm which
> fails unless lvm was compiled in (yes, it was docmented but it
> should probably be a tad smarter ;-) ).
> 
> # modprobe lvm-mod
> 
> (I might have had to do a vgscan & vgchange -a y)
> 
> # mount -a
> 
> so here is where I was looking for help... I get an I/O error
> but things come up:
> 
> # vgck -v
> vgck -- locking logical volume manager
> vgck -- finding all volume group(s)
> vgck -- checking volume group name "vg1"
> vgck -- checking existence of volume group "vg1"
> vgck -- reading volume group data for "vg1" from lvmtab
> vgck -- checking volume group consistency  of "vg1" in lvmtab
> vgck -- VGDA of "vg1" in lvmtab is consistent
> vgck -- reading volume group data for "vg1" from physical volume(s)
>  I/O error: dev 08:11, sector 0
> vgck -- checking volume group consistency  of "vg1" on physical volume
> vgck -- VGDA of "vg1" on physical volume is consistent
> vgck -- unlocking logical volume manager
> 
> Is this a problem and how do I fix it?  This problem did not exist
> before I rebooted.
> 
> Thanks in advance,
> 
> 
> /Allan
> -- 
> Allan M. Wind			email: allanwind@mediaone.net
> P.O. Box 2022			finger: awind@digit-safe.dyndns.org (GPG/PGP)
> Woburn, MA 01888-0022
> USA

*** Software bugs are stupid.
    Nevertheless it needs not so stupid people to solve them ***

=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-

Heinz Mauelshagen                                 Sistina Software Inc.
Senior Consultant/Developer                       Am Sonnenhang 11
                                                  56242 Marienrachdorf
                                                  Germany
Mauelshagen@Sistina.com                           +49 2626 141200
                                                       FAX 924446
=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-

Re: [linux-lvm] I/O error: dev 08:11, sector 0

[Patrick Caulfield]

On Thu, Jul 19, 2001 at 11:13:51AM +0200, Heinz J. Mauelshagen wrote:
> 
> Allan,
> 
> 08:11 (/dev/sda11) shows that I/O error, you use /dev/sdb, which is 08:16
> for your VG -> no problem for LVM.
> 
> Because /dev/sda11 seems to be configured, LVM reads it on its search for
> physical volumes.
> 
> If you are not using /dev/sda11 for anything you are fine for the moment.

If the minor number is in hex that makes it /dev/sdb1 - which seems slightly
more sensible to me :-)

patrick

Re: [linux-lvm] I/O error: dev 08:11, sector 0

[Allan M. Wind]

[-- Attachment #1: Type: text/plain, Size: 1177 bytes --]

On 2001-07-19 11:13:51, Heinz J. Mauelshagen wrote:

> 08:11 (/dev/sda11) shows that I/O error, you use /dev/sdb, which is 08:16
> for your VG -> no problem for LVM.
> If there's a file system on /dev/sda11 *and* just the first sector is not
> accessable, you can still happily access the filesystem, because it doesn't use
> that sector at all.
> But you should check and probably replace your sda drive to avoid
> trouble in the future.

Hmm... I do not have a /dev/sda11, only sda1 & sda2 (root and swap
respectively).  However, I was aware that my sdc drive was dying
(the drive was scsi id 8 / sdb when this was logged):

Jul 14 09:33:58 digit-safe kernel: scsi0: ERROR on channel 0, id 8, lun 0, CDB: Read (10) 00 00 09 c9 48 00 00 18 00 
Jul 14 09:33:58 digit-safe kernel: Info fld=0x9c952, Current sd08:12: sense key Medium Error
Jul 14 09:33:58 digit-safe kernel: Additional sense indicates Data synchronization mark error
Jul 14 09:33:58 digit-safe kernel:  I/O error: dev 08:12, sector 577872


/Allan
-- 
Allan M. Wind			email: allanwind@mediaone.net
P.O. Box 2022			finger: awind@digit-safe.dyndns.org (GPG/PGP)
Woburn, MA 01888-0022
USA

[-- Attachment #2: Type: application/pgp-signature, Size: 232 bytes --]

Re: [linux-lvm] I/O error: dev 08:11, sector 0

[Heinz J. Mauelshagen]

On Thu, Jul 19, 2001 at 01:48:47PM +0100, Patrick Caulfield wrote:
> On Thu, Jul 19, 2001 at 11:13:51AM +0200, Heinz J. Mauelshagen wrote:
> > 
> > Allan,
> > 
> > 08:11 (/dev/sda11) shows that I/O error, you use /dev/sdb, which is 08:16
> > for your VG -> no problem for LVM.
> > 
> > Because /dev/sda11 seems to be configured, LVM reads it on its search for
> > physical volumes.
> > 
> > If you are not using /dev/sda11 for anything you are fine for the moment.
> 
> If the minor number is in hex that makes it /dev/sdb1 - which seems slightly
> more sensible to me :-)

Yes, you are right. Got that base wrong :)

> 
> patrick
> 
> _______________________________________________
> linux-lvm mailing list
> linux-lvm@sistina.com
> http://lists.sistina.com/mailman/listinfo/linux-lvm
> read the LVM HOW-TO at http://www.sistina.com/lvm/Pages/howto.html

Regards,
Heinz    -- The LVM Guy --

*** Software bugs are stupid.
    Nevertheless it needs not so stupid people to solve them ***

=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-

Heinz Mauelshagen                                 Sistina Software Inc.
Senior Consultant/Developer                       Am Sonnenhang 11
                                                  56242 Marienrachdorf
                                                  Germany
Mauelshagen@Sistina.com                           +49 2626 141200
                                                       FAX 924446
=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-

Re: [linux-lvm] I/O error: dev 08:11, sector 0

[Heinz J. Mauelshagen]

Allan,

as Patrick pointed out, I got the base wrong.

Are you sure that the drive is dying? 

You might have one of those nasty SCSI cable length/termination/power problems
causing this. Happens quite often after a configuration change on a SCSI bus.

Heinz

On Thu, Jul 19, 2001 at 09:10:14AM -0400, Allan M. Wind wrote:
> On 2001-07-19 11:13:51, Heinz J. Mauelshagen wrote:
> 
> > 08:11 (/dev/sda11) shows that I/O error, you use /dev/sdb, which is 08:16
> > for your VG -> no problem for LVM.
> > If there's a file system on /dev/sda11 *and* just the first sector is not
> > accessable, you can still happily access the filesystem, because it doesn't use
> > that sector at all.
> > But you should check and probably replace your sda drive to avoid
> > trouble in the future.
> 
> Hmm... I do not have a /dev/sda11, only sda1 & sda2 (root and swap
> respectively).  However, I was aware that my sdc drive was dying
> (the drive was scsi id 8 / sdb when this was logged):
> 
> Jul 14 09:33:58 digit-safe kernel: scsi0: ERROR on channel 0, id 8, lun 0, CDB: Read (10) 00 00 09 c9 48 00 00 18 00 
> Jul 14 09:33:58 digit-safe kernel: Info fld=0x9c952, Current sd08:12: sense key Medium Error
> Jul 14 09:33:58 digit-safe kernel: Additional sense indicates Data synchronization mark error
> Jul 14 09:33:58 digit-safe kernel:  I/O error: dev 08:12, sector 577872
> 
> 
> /Allan
> -- 
> Allan M. Wind			email: allanwind@mediaone.net
> P.O. Box 2022			finger: awind@digit-safe.dyndns.org (GPG/PGP)
> Woburn, MA 01888-0022
> USA

=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-

Heinz Mauelshagen                                 Sistina Software Inc.
Senior Consultant/Developer                       Am Sonnenhang 11
                                                  56242 Marienrachdorf
                                                  Germany
Mauelshagen@Sistina.com                           +49 2626 141200
                                                       FAX 924446
=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-

Re: [linux-lvm] I/O error: dev 08:11, sector 0

[Allan M. Wind]

[-- Attachment #1: Type: text/plain, Size: 1127 bytes --]

On 2001-07-19 16:06:07, Heinz J. Mauelshagen wrote:

> as Patrick pointed out, I got the base wrong.
> 
> Are you sure that the drive is dying? 
> 
> You might have one of those nasty SCSI cable length/termination/power problems
> causing this. Happens quite often after a configuration change on a SCSI bus.

The drive started complaining before any changes were made, I suppose the cable
could have gone bad (length, termination, power etc should be ok), but...

When does lvm create /dev/sdb1 (I only used /dev/sdb)?

The thing that I find fishy about the hardware problem idea is that (1) vgck did
not see anything wrong before the reboot and (2) I went through the exercise twice 
(outlined earlier) and got the _same_ I/O error upon reboot.

Supporting evidence for (non-drive) hardware issues is of course that now my new
sdb drive appears effected in addition to sdc which reported media failures earlier.

I really appreciate the help, btw,


/Allan
-- 
Allan M. Wind			email: allanwind@mediaone.net
P.O. Box 2022			finger: awind@digit-safe.dyndns.org (GPG/PGP)
Woburn, MA 01888-0022
USA

[-- Attachment #2: Type: application/pgp-signature, Size: 232 bytes --]

Re: [linux-lvm] I/O error: dev 08:11, sector 0

[Allan M. Wind]

[-- Attachment #1: Type: text/plain, Size: 1275 bytes --]

On 2001-07-19 10:52:15, Allan M. Wind wrote:

> The drive started complaining before any changes were made, I suppose the cable
> could have gone bad (length, termination, power etc should be ok), but...

Replaced the scsi cable per suggestion and it appears to make no
difference in that 'I/O error: dev 08:11, sector 0' is logged when
running vgck.  No "scsi:" errors were reported with this drive.

> When does lvm create /dev/sdb1 (I only used /dev/sdb)?

Still curios about this one as I did not create it and cfdisk does not
seem to recognize the partition (just reports free space).  How is
this partition locatated in relation to the disc device?  I get the
above error if dd if=sdb1 but I would think the sector would be
reachable via the the disc device also... perhaps there is something
wrong with the device partition?

Does lvm use the first sector?  If not, could it just be bad data that
is somehow read and therefor complained about and could be wiped?
Should I create a partition and feed that to pvcreate instead of
giving it the raw disk?

Any input is greatly appreciated...


/Allan
-- 
Allan M. Wind			email: allanwind@mediaone.net
P.O. Box 2022			finger: awind@digit-safe.dyndns.org (GPG/PGP)
Woburn, MA 01888-0022
USA

[-- Attachment #2: Type: application/pgp-signature, Size: 232 bytes --]

Re: [linux-lvm] I/O error: dev 08:11, sector 0

[Allan M. Wind]

[-- Attachment #1: Type: text/plain, Size: 1163 bytes --]

On 2001-07-20 21:13:56, Allan M. Wind wrote:

> > When does lvm create /dev/sdb1 (I only used /dev/sdb)?
> 
> Still curios about this one as I did not create it and cfdisk does not
> seem to recognize the partition (just reports free space).  How is
> this partition locatated in relation to the disc device?  I get the
> above error if dd if=sdb1 but I would think the sector would be
> reachable via the the disc device also... perhaps there is something
> wrong with the device partition?
> 
> Does lvm use the first sector?  If not, could it just be bad data that
> is somehow read and therefor complained about and could be wiped?
> Should I create a partition and feed that to pvcreate instead of
> giving it the raw disk?

Used `pvcreate /dev/sdb1` (instead of /dev/sdb) and everything works
as expected.  I suspect there is a bug with the combination of linux
2.4.6 (devfsd) and lvm 0.9.1beta7 causing a non-functional partition
to be created when you use the raw disk partition.


/Allan
-- 
Allan M. Wind			email: allanwind@mediaone.net
P.O. Box 2022			finger: awind@digit-safe.dyndns.org (GPG/PGP)
Woburn, MA 01888-0022
USA

[-- Attachment #2: Type: application/pgp-signature, Size: 232 bytes --]

Re: [linux-lvm] I/O error: dev 08:11, sector 0

[Heinz J. Mauelshagen]

On Sat, Jul 21, 2001 at 01:44:19PM -0400, Allan M. Wind wrote:
> On 2001-07-20 21:13:56, Allan M. Wind wrote:
> 
> > > When does lvm create /dev/sdb1 (I only used /dev/sdb)?
> > 
> > Still curios about this one as I did not create it and cfdisk does not
> > seem to recognize the partition (just reports free space).  How is
> > this partition locatated in relation to the disc device?  I get the
> > above error if dd if=sdb1 but I would think the sector would be
> > reachable via the the disc device also... perhaps there is something
> > wrong with the device partition?
> > 
> > Does lvm use the first sector?  If not, could it just be bad data that
> > is somehow read and therefor complained about and could be wiped?
> > Should I create a partition and feed that to pvcreate instead of
> > giving it the raw disk?
> 
> Used `pvcreate /dev/sdb1` (instead of /dev/sdb) and everything works
> as expected.  I suspect there is a bug with the combination of linux
> 2.4.6 (devfsd) and lvm 0.9.1beta7 causing a non-functional partition
> to be created when you use the raw disk partition.

Are you able to reporduce this?

> 
> 
> /Allan
> -- 
> Allan M. Wind			email: allanwind@mediaone.net
> P.O. Box 2022			finger: awind@digit-safe.dyndns.org (GPG/PGP)
> Woburn, MA 01888-0022
> USA



-- 

Regards,
Heinz    -- The LVM Guy --

*** Software bugs are stupid.
    Nevertheless it needs not so stupid people to solve them ***

=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-

Heinz Mauelshagen                                 Sistina Software Inc.
Senior Consultant/Developer                       Am Sonnenhang 11
                                                  56242 Marienrachdorf
                                                  Germany
Mauelshagen@Sistina.com                           +49 2626 141200
                                                       FAX 924446
=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-

Re: [linux-lvm] I/O error: dev 08:11, sector 0

[Allan M. Wind]

[-- Attachment #1: Type: text/plain, Size: 752 bytes --]

On 2001-07-23 10:02:03, Heinz J. Mauelshagen wrote:

> > Used `pvcreate /dev/sdb1` (instead of /dev/sdb) and everything works
> > as expected.  I suspect there is a bug with the combination of linux
> > 2.4.6 (devfsd) and lvm 0.9.1beta7 causing a non-functional partition
> > to be created when you use the raw disk partition.
> 
> Are you able to reporduce this?

Yes, in the sense that I managed to get into this condition twice using the
steps outlined earlier in the thread.  It will be a bit painful for me
to provide more data at this point, but I will backup/restore if needed.


/Allan
-- 
Allan M. Wind			email: allanwind@mediaone.net
P.O. Box 2022			finger: awind@digit-safe.dyndns.org (GPG/PGP)
Woburn, MA 01888-0022
USA

[-- Attachment #2: Type: application/pgp-signature, Size: 232 bytes --]