pvcreate Warning: couldn't query physical volume

[Richard Plourde_3]

I have a strange one here.

I have two systems connecting to a StorageWorks DS2405 to JBODs. Setup in a Private Loop, both see the same disks.

I setup volume group tsdm1a on system A, vexport -p -m, and tried to import it on system B. It failed so I stop at this point.

Later, I had to reboot system A. Now I can't do a vgchange on tsdm1a. I tried to do a vgcfgrestore, no success.

I managed to find the VGID on the pv it supposedly can't query and inserted it in my tsdm1a.map. I managed to import my VG but I still get the same error with the vgchange:

tspmtl01:[/usr/contrib/bin]#vgchange -a y tsdm1a
vgchange: Warning: couldn't query physical volume "/dev/dsk/c6t2d0":
The specified path does not correspond to physical volume attached to
this volume group
vgchange: Warning: couldn't query physical volume "/dev/dsk/c7t2d0":
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 "tsdm1a":
Quorum not present, or some physical volume(s) are missing.

Pvdisplay of the pv give this:

tspmtl01:[/usr/contrib/bin]#pvdisplay -v /dev/dsk/c6t2d0
pvdisplay: Warning: couldn't query physical volume "/dev/dsk/c6t2d0":
The specified path does not correspond to physical volume attached to
this volume group
pvdisplay: Warning: couldn't query physical volume "/dev/dsk/c7t2d0":
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/tsdm1a".
pvdisplay: Cannot display physical volume "/dev/dsk/c6t2d0".
You have mail in /var/mail/root
tspmtl01:[/usr/contrib/bin]#

As you might have guessed, c6t2d0 and c7t2d0 are the same disk (seen via two paths).

I can see the disk through ioscan as CLAIMED. The "dd to null" command works fine on this disk.
I am running on two L3000 with HP-UX 11.0.

[Sandman!]

What does "strings /etc/lvmtab" show. You might have to recrate that file. Save the exisiting and create a new one by running "vgscan -a -v". See if that fixes it.

[Richard Plourde_3]

Yeah, sorry, I did try to recreate the lvmtab file. It didn't work.

I do see the proper devices for my vg when I strings the lvmtab file. Here is an excerpt:

tspmtl01:[/var/VRTSvcs/log]#strings /etc/lvmtab
/dev/vg00
...
...
...
/dev/tsdm1a
/dev/dsk/c6t2d0
/dev/dsk/c7t2d0
...
...
...

[Hoang Chi Cong_1]

Hi

In your case, may be the path has problem when the volume group doesn't active.

To solve it, just active the pv again:
#pvchange -a y /dev/dsk/c6t2d0

Then you can active the volume group.
#vgchange -a y tsdm1a

Once more:

http://forums1.itrc.hp.com/service/forums/publicProfile.do?userId=CA1163800&forumId=1

You should assign some points to who takes time help you

Thanks

[Hoang Chi Cong_1]

forgot it...
Please show me the result...

Goodluck!

[Mridul Shrivastava]

what is the error message if you execute the vgcfgrestore ??

Please check do we have valid configuration backup under /etc/lvmconf

[Srinivas Thokala_1]

When you've noticed that there is a problem with the map files, before rebooting the system you would've verified for any instances got failed on this volume with vgdisplay and lvdisplay in verbose mode.

Since you've already rebooted the server and if you have preserved the mapfiles for /dev/tsdm1a, I would say try creating another test vg and try importing to the new one. If you have preserved any old dated map files try using them.

Before you even try this make an attempt to
#vgchange -a n /dev/tsdm1a
For exporting mapfiles use: #vgexport
-p -s -m /dev/tsdm1as.map /dev/tsdm1a
and for importing use #vgimport
-s -m /dev/tsdm1as.map /dev/tsdm1a
#vgchange -a y /dev/tsdm1a
#strings /etc/lvmtab

-Srini