vgextend for SG cluster lock disk

[Stuart Powell]

When attempting to extend vg01 to include the future ServiceGuard cluster lock disk I get the following error message:
#vgextend /dev/vg01 /dev/dsk/c15t1d0
vgextend: Couldn't install the physical volume "/dev/dsk/c15t1d0".
No space left on device

I preceeded this command with pvcreate /dev/rdsk/c15t1d0.

Any ideas?

Stuart

[Ian Dennison_1]

Does a 'pvdisplay' or 'diskinfo' show size on the disk?

Also, does a 'strings /etc/lvmtab |greop c15t1d0' show anything?

I always use 'pvcreate -f' (force) to make doubly sure. Have you tried that?

Share and Enjoy! Ian

[Stuart Powell]

#diskinfo /dev/rdsk/c15t1d0
SCSI describe of /dev/rdsk/c15t1d0:
vendor: DGC
product id: C4700WDR5
type: direct access
size: 4096 Kbytes
bytes per sector: 512
Then I took your advice:
#pvcreate -f /dev/rdsk/c15t1d0
Physical volume "/dev/rdsk/c15t1d0" has been successfully created.
#vgextend /dev/vg01 /dev/dsk/c15t1d0
Current path "/dev/dsk/c13t1d1" is an alternate link, skip.
Current path "/dev/dsk/c15t1d1" is an alternate link, skip.
vgextend: Couldn't install the physical volume "/dev/dsk/c15t1d0".
No space left on device

Stuart

[Uday_S_Ankolekar]

How much is max_pe is set while creating vg01??

If you are already reached the limit then you can not extend your vg01 any more.

You have to set this limit while creating VG. Try recreating new VG

-USA..

[Sridhar Bhaskarla]

Hi Stuart,

Do a bdf and see if any of the file systems is at 100%. Particularly /. Look at your /var/adm/syslog/syslog.log for more messages with "no space".

-Sri

[Stuart Powell]

USA - Max PE per PV is 62500 (I've run into that problem before, so I create this one big) Max PV is 16, and Cur PV is 2.

Sir - Nothing is at 100% on any filesystem. However the database space on vg02 is all used up.

I'm leaving for the day, but would appreciate any other input.

Stuart

[James R. Ferguson]

Hi Stuart:

I'm wondering, if you are running out of space in the root directory for the 'lvmconf' file, especially since your LVM data is going to be very large.

Try the 'vgextend' *without* concommitant creation of the '/etc/lvmconf/vg01.conf' file:

# vgextend -A n /dev/vg01 /dev/dsk/c15t1d0

Regards!

...JRF...

[Mladen Despic]

Have you checked your syslog?
There may be a kernel parameter issue. Perhaps you are running out of the inodes? Check this with 'bdf -i'

[Pradeep_3]

Is disk o.k.? Try creatiing new VG using that disk.

Paresh

[Armin Feller]

Hi,

are the current recommended LVM, SCSI, FS patches installed? Is there no full filesystem (maybe a full root filesystem can cause such a message)? Is the volume group active on the node you try to extend it? Is the minor number unique? Is there a alternative path to the disk you are using instead of the primary one? ...?

Please answer all this questions, hope that helps to fix the problem.

Regards ...
Armin

[Stuart Powell]

Thanks to everyone for your input. I gave up and called HP's Response Center. We got to the second line before the engineer asked the critical questions: How big is the LUN? What is the PE size for the VG? The answer to the first question was 4Mb and the answer to the second was 8Mb. A physical device needs to be at least 1 PE. I've got to recreate my cluster lock LUN's to be 8Mb.

Stuart