Volume Group Creation

[Paul McGahan]

I have four discs which I want to put into a volume group.

I have initialised all the discs using pvcreate and created the volume group directory and group file.

When I create the volume group with the first disk using the command:
vgcreate /dev/vg_ecs /dev/dsk/c1t4d0
the volume group gets created but I also get the error message 'VGRA appears corrupted'

No backup configuration file is created and when I run vgcfgbackup manually I again get the message VGRA appears corrupted.

I have written a file to the disk using the 'dd' command in order to erase any previous volume group information and then reinitialised it. However this does not seem to cure the problem.

Has anyone seen this problem before ?

[John Palmer]

Hi Paul,

No I've not seen this before. Have you tried creating the volume group with any combination of the other three disks?

This would eliminate that particular disk from being a problem.

Regards,

John

[CHRIS_ANORUO]

Instead of call the device vg_ecs, name it /dev/vg1ecs.
1.First do pvcreate /dev/rdsk/c1t4d0(Initialize the disk)
2.mkdir /dev/vg1ecs(create a directory for vol group)
3.mknod /dev/vg1ecs/group c 64 0xNN000(create a device file named group)
4.vgcreate /dev/vg1ecs /dev/dsk/c1t4d0 (create volume group)
5.lvcreate /dev/vg1ecs(to create logical volume)

[John Palmer]

I did initially suspect the underscore in your volume group name as I had never seen it used before but I have successfully created a volume group with the same name on a 10.20 server. I don't have one at 11.00 with a spare disk unfortunately.

[Cheryl Griffin · Accepted solution]

Paul,

What version of the operating system are you using? There was an issue at 10.20 that was addressed by the latest LVM cumulative patch.
Patch Name: PHCO_18563
Patch Description: s700_800 10.20 LVM commands cumulative patch

This patch has other LVM related patch dependencies which must be installed at the same time. (If you need the list, I can post it here or email it to you.)

Best Wishes,
Cheryl

[Vincente Fernandes]

May be vgcreate doesn't like "_" in vgname. name it /dev/vgecs.
1.Do pvcreate on all 4 disks
2.mkdir /dev/vg1ecs
3.mknod /dev/vg1ecs/group c 64 0xNN000
4.vgcreate /dev/vg1ecs on all 4 disks
or vgcreate on 1st disk
5.create all the logical volumes.
6.do vgextend /dev/vgecs on all the 3 disks( if vgcreate was done only on 1st disk in step 4)
Will u be mirroring these disks?

[Antoanetta Naghiu]

I'm not sure this helps, but what happen if you try to force pvcreate -f /dev/rdsk/... and after that try to see it with pvdisplay -v.
What type of disk do you use?
As already said above, see if you have the LVM patches.

[Stefan Farrelly]

The timestamps on the disk have become so large that they become negative when it is treated as signed integer (since HP-UX 10.10), thus VG commands fail due to a negative timestamp. I take it these disks are old and have been used a bit ?

Try pvcreate - f /dev/rdsk/cxxxxx first then add it to a VG, if you still get the error use mediainit on the disk first, then you should be able to add it into a VG no problem.

[Account Not Used]

The underscore "_" is not the problem. You may have a bad disk. I would try using the
-f option when creating the physical volume.
EXAMPLE: pvcreate -f /dev/rdsk/cxtxdx.Also I think someone else mentioned above as to when the error apears.Adding one at a time should tell you what disk is bad. The -f will force the creation and delete any previous filesystem that may have been on it prior. There is also the -f option in the vgcreate as well. Let us know how it works out. If you use the -f option for vgcreate make usre the physical extent size are all the same size or you can cause corruption.

[Sundararajan Anandhan]

Paul,

I have seen the same problem with Hpux 10.20
and fixed it after loading few patches as suggested by Cheryl.
(Load the dependencies as well.)

Good Luck,
Sundar

[Michael F. Steele]

mediainit the disk in question to erase everything on the disk. Then pvcreate.