topic Re: vgimport error in Operating System - HP-UX

vgimport error

Paolo Bardi — Wed, 05 Apr 2000 07:52:46 GMT

Hi, I have a little problem with "vgimport".

I have two system in service guard with a AutoRaid
mass storage.
I have configured other 3 LUN and the system created 3 new physical volume
(/dev/dsk/xxxxx) where I created 3 new filesystem.
After this I did this command:

system1: #vgexport -p -s -v MAPFILE /dev/dsk/vgxxx

transferred MAPFILE into system2

system2: recreated the vgxxx
system2: #vgimport -s -m MAPFILE /dev/vgxxx

After this

vgimport: Physical volume "/dev/dsk/cxtxdx" is not a block special file.
....
....
....

Why???

Many Thanks in advances.

Paolo

Re: vgimport error

Mitch Wolf — Wed, 05 Apr 2000 08:29:59 GMT

Hi

Looks backwards should be

vgexport -p -s -m MAPFILE /dev/vg##

vgimport -s -m MAPFILE /dev/vg## /dev/dsk/c#t#d#

Mitch

Re: vgimport error

Fernando Santana — Thu, 06 Apr 2000 08:48:58 GMT

Hi, you first have to know that there is a connection between the second
system and the disks.
Run "ioscan -fnC disk" to verify the disks are there and run "#insf -e" to have
the device special files
installed. I believe you have to name the disks after
the vgimport command if they are seen on different hardware paths from that of
the first system.

vgimport -m MAPFILE /dev/vg?? /dev/dsk/c?t?d? ....

Good Luck.

Re: vgimport error

Bernadette Keohan — Wed, 07 Jun 2000 04:02:28 GMT

It's my understanding that by using the "-s" option on the vgexport command you no longer have to specify the actual physical devices on the command line when importing.

On export the "-s" option places the VGID header into the mapfile. When importing the "-s" option causes the import command to search all physical volume headers to find devices matching the VGID listed in the mapfile. Once found it automatically imports them. This is very beneficial when you have multiple disks. This option is not recommended however when using "EMC".

Hope this helps,
Bernie

Re: vgimport error

Alan Riggs — Wed, 07 Jun 2000 13:28:28 GMT

It appears to me that your cluster does not address the autoraid consistently. That is, the device files for the disks are not identical accross machines. Run ioscan on each box and make sure the SCSI paths are the same. If they are not, then you will probably want to rearrange the order in which the system discovers devices by utilizing ioinit. (man ioinit for procedure.)

You will need to recreate your disk device files (using insf) after ioscan returns your devices in the expected order. You may, however, want to check your cabling and ensure that you have the autoraid connected identically to each machine. This is not an absolute requirement, but it will make your life easier.