Operating System - HP-UX
1753723 Members
4772 Online
108799 Solutions
New Discussion

Problem with vgextend command

 
Toscanini Montaño
Occasional Advisor

Problem with vgextend command

I have some new LUNs created in VA7100 and they are shared in two HP-UX 11.11 server and controlled by service guard.

I made the extension of the VG with the new LUNs on the primary HP-UX server but I can’t extend the same VG on the second HP-UX server.

When I run the vgextend command the system return the following message:

vgextend /dev/vgdbofprod /dev/dsk/c10t3d4
Current path "/dev/dsk/c10t0d2" is an alternate link, skip.
Current path "/dev/dsk/c8t1d0" is an alternate link, skip.
vgextend: Couldn't install the physical volume "/dev/dsk/c10t3d4".
Device busy

Please, could somebody help me?

Thanks
Toscanini
8 REPLIES 8
Michal Kapalka (mikap)
Honored Contributor

Re: Problem with vgextend command

hi,

you need to shutdown the cluster export the eg from node 1 and them import it to the node 2.


mikap
Toscanini Montaño
Occasional Advisor

Re: Problem with vgextend command

thanks Mikap, i will try it asap.

Toscanini.-
smatador
Honored Contributor

Re: Problem with vgextend command

Hi,
If the disk is on another vg, it have a vgid of the another vg. So it's normal that it's busy. So deactivate the vg on the first node and distribute the volume group with vgexport/vgimport.
Hope it helps
Michael Steele_2
Honored Contributor

Re: Problem with vgextend command

I've been trying to upload a MC/SG manual for you but it 3 MB and over the max limit of 1 MB. Sorry. But, you should stop and find a manual. There are many steps to this complicated procedure, one for sure is the following. For you're problem, you vgexport a map file from node one and ftp it over to node two, where you vgimport the map file and create a duplicate vg. But, in short, you are missing many steps and should call the response center and search the ITRC for a manual.
Support Fatherhood - Stop Family Law
wci
Frequent Advisor

Re: Problem with vgextend command

Hi

You can find the manual here.

http://docs.hp.com/en/ha.html#Serviceguard

or this page onwards
http://docs.hp.com/en/B3936-90079/ch05s04.html

Regards

WCI

Anoop P_2
Regular Advisor

Re: Problem with vgextend command

Hello,

You do not need to perform extension on the second server. Infact you should not attempt to.

When you extended it on the primary server, it already modified the LVM headers on the disks with the new information. Now the disks being shared and viewable by both servers in the cluster (as it should be), you only need to update the lvmtab and create relevant device files for the new luns, on the secondary server.

Follow below:

perform ioscan and ensure disks are claimed
#insf -e #Creates device files

On Primary server:

For each vg that that you added luns to, that is shared and active on primary server, do

#vgexport -p -v -s -m #Creates map file with all lv info

Copy this map file to the secondary server
On Secondary server:

Note down the minor numbers of the group file

#ll /dev/vg*/group > tmpfile

For each vg that you added luns to, shared, and that is active now on primary server, do

#vgexport vgname # Exports and removes configuration
#mkdir /dev/vgname
#mknod /dev/vgname/group c 64 # This creates the device file with the same minor number as before.
#vgimport -s -v -m # This will scan all the available disks, match them with the vgid, import with relevant pvs and create device files for lvols.

Remember that a different map file needs to be used for each vg.

And all this can be performed WITHOUT down time when the cluster and packages are up and running :)


nightwich
Valued Contributor

Re: Problem with vgextend command

Hi Toscanini


You should follow the correct approach described by "Anoop P".

This is the correct order to do what you are attended to do ..

If you have any problem please reply.

Regards.
kevin_m
Valued Contributor

Re: Problem with vgextend command

If you created any new logical volumes on these LUNs you'll want to update the package control scripts (on all nodes) to automatically mount/unmount the new file systems.
- Kevin