VG Activation Problem

[Karthik S S]

Hi,
In one of our system we have a VG named vg06. Today when I tried to activate that it is giving the following error,


# vgchange -a y vg06
vgchange: Warning: Couldn't attach to the volume group physical volume "/dev/dsk
/c12t2d2":
Cross-device link
vgchange: Warning: Couldn't attach to the volume group physical volume "/dev/dsk
/c12t2d3":
Cross-device link
vgchange: Warning: couldn't query physical volume "/dev/dsk/c12t2d2":
The specified path does not correspond to physical volume attached to
this volume group
vgchange: Warning: couldn't query physical volume "/dev/dsk/c12t2d3":
The specified path does not correspond to physical volume attached to
this volume group
vgchange: Warning: couldn't query all of the physical volumes.
vgchange: Couldn't activate volume group "vg06":
Quorum not present, or some physical volume(s) are missing.


The output of strings command as follows,
# strings /etc/lvmtab
/dev/vg00
/dev/dsk/c12t5d7
/dev/vg06
/dev/dsk/c12t2d2
/dev/dsk/c12t2d3

Both the disks are present and accessible from the system using diskinfo command.

# diskinfo /dev/rdsk/c12t2d2
SCSI describe of /dev/rdsk/c12t2d2:
vendor: HP
product id: OPEN-8
type: direct access
size: 7175520 Kbytes
bytes per sector: 512

# diskinfo /dev/rdsk/c12t2d3
SCSI describe of /dev/rdsk/c12t2d3:
vendor: HP
product id: OPEN-8
type: direct access
size: 7175520 Kbytes
bytes per sector: 512

What could be the problem? Pl. let me know

Thanks
Karthik

[Jerome Baron · Accepted solution]

Hi,

You are getting the message because the /dev/vgxx/group device file has duplicate minor numbers.

Perform the following steps to resolve the error:

1. vgexport vgxx

2. Re-import the mapfile.

Regards,
Jerome

[Michael Tully]

Make sure that you can actually talk to these disks.

# dd if=/dev/rdsk/c12t3d0 of=/dev/null bs=1024

If you can't actually talk to the disk (get an IO error) you may have a hardware problem like a controller.

[Elif Gius]

Hi,

first try an ioscan -fnC disk to see if the disks are claimed...
I had it sometimes that diskinfo doesn't give the correct information because it remembers only the infos from the last boot time...

and check like suggested the minor numbers...maybe another vg was created with the same number

ls -la /dev/vg*/group

what does pvdisplay say ???

[Karthik S S]

Hi all,

Thank you for the suggestions. But as such the minor numbers of the VGs configured in the system are unique. However there are some directories of OLD VGs in under the /dev directory. For instance there is a vglock directory which contains the group files and lvols with same minor number as VG06. However that VG is not listed in the /etc/lvmtab file.

The output of pvdisplay for the Physical disks corresponding to the VG06 as follows,


# pvdisplay /dev/dsk/c12t2d2
pvdisplay: Warning: couldn't query physical volume "/dev/dsk/c12t2d2":
The specified path does not correspond to physical volume attached to
this volume group
pvdisplay: Warning: couldn't query physical volume "/dev/dsk/c12t2d3":
The specified path does not correspond to physical volume attached to
this volume group
pvdisplay: Warning: couldn't query all of the physical volumes.
pvdisplay: Couldn't retrieve the names of the physical volumes
belonging to volume group "/dev/vg06".
pvdisplay: Cannot display physical volume "/dev/dsk/c12t2d2".
#

# pvdisplay /dev/dsk/c12t2d3
pvdisplay: Warning: couldn't query physical volume "/dev/dsk/c12t2d2":
The specified path does not correspond to physical volume attached to
this volume group
pvdisplay: Warning: couldn't query physical volume "/dev/dsk/c12t2d3":
The specified path does not correspond to physical volume attached to
this volume group
pvdisplay: Warning: couldn't query all of the physical volumes.
pvdisplay: Couldn't retrieve the names of the physical volumes
belonging to volume group "/dev/vg06".
pvdisplay: Cannot display physical volume "/dev/dsk/c12t2d3".
#

Any idea?

However I have not tried vgexport and vgimport since I have not got the permission from our client.

Thanks
Karthik

[Jerome Baron]

Hi,

Your problem seems cause by a bad disk with HW problem. I suggest you change this disk.
What does ioscan say about this disk ?

Regards,
Jerome

[Jerome Baron]

Re,

You can confirm a HW problem with :
dd if=/dev/rdsk/cxtydz of=/dev/null

Regards,
Jerome

[Karthik S S]

Hi Jerome,

The ioscan shows the disks as CLAIMED,

#ioscan -fnC disk
disk 33 0/2/0/0.8.0.31.0.2.2 sdisk CLAIMED DEVICE HP OPEN-8
/dev/dsk/c12t2d2 /dev/rdsk/c12t2d2
disk 34 0/2/0/0.8.0.31.0.2.3 sdisk CLAIMED DEVICE HP OPEN-8
/dev/dsk/c12t2d3 /dev/rdsk/c12t2d3


Rightnow I am trying a dd on these disks and they are still running. So far I haven't received any I/O errors. Let you know once it is complete ...

Thanks
Karthik

[Jerome Baron]

Karthik,

Your device is a XP LUN, is it any modification done on the array or on the lun configuration ?

Regards,
Jerome

[Karthik S S]

Hi

No changes have been made in XP LUN Configuration. How much time would dd'ing a 7GB of volume will take. I have started before 45min. still it is not over.

Thanks
Karthik

[Jerome Baron]

Karthik,

I don't know how long does the dd run for 7gb.
Can you check if you have message in syslog ?

Regards,
Jerome

[Elif Gius]

You can interruppt dd...then you will get a statistic how much was dd.

what does this say ???


can you delete the old vglock directory ???

[Karthik S S]

Elif,

So far no errors I think there is no problem with the disks,

# dd if=/dev/rdsk/c12t2d2 of=/dev/null
4564832+0 records in
4564832+0 records out
#

# dd if=/dev/rdsk/c12t2d3 of=/dev/null
4824610+0 records in
4824610+0 records out
#

And I have tried moving the vglock directory to a different location. But still I am unable to activate vg06. :-(

Karthik

[T G Manikandan]

Hello
I assume that the vg info was lost.

Can you do a
#vgcfgrestore -n /dev/vg06 /dev/rdsk/c12t2d2
#vgcfgrestore -n /dev/vg06 /dev/rdsk/c12t2d3
check for the man pages of the vgcfgrestore

Thanks

[Elif Gius]

So it should not be HW Problem ...

Try like suggested vgcfgrestore ...


or try moving /etc/lvmtab to /etc/lvmtab.old

and then vgscan ...

[Pete Randall]

Just a tip on your dd taking so long - use a large block size. You didn't specify anything so dd defaults to 512 byte blocks. For a 7GB lun, I would guess that could take a week or so. In the future, try something like this:

dd if=/dev/rdsk/c12t2d2 of=/dev/null bs=1024k

As for your VG problem, I'd try the vgscan idea. Move /etc/lvmtab first then run vgscan and see what you get.

Pete

[Karthik S S]

Hi

I have suggested our Client to try vgcfgrestore. I will let you know the results once they get back to me. Thanks a ton for all your inputs.

Thanks
Karthik

[John Palmer]

I'd still be concerned about what happened to these LUNS. They've obviously lost the LVM configuration data (which you're going to restore with vgcfgrestore) so expect them to have lost all your logical volume data as well.

I tend to agree with Jerome that there's a chance that the array has been reconfigured. Another possibility is that the LUNS have been overwritten by another server. Do you have more than one server attached to the XP?

I hope you've got a good backup tape!

Regards,
John