HP-UX ServiceGuard Secondary Node LVM Awareness

[coke]

Forum Post: HP-UX ServiceGuard Secondary Node LVM Awareness Problem Statement:

After expanding storage and extending LVM on ServiceGuard primary node, how to ensure secondary node recognizes the changes for proper failover?

Scenario:

• Storage admin expanded LUNs from 100GB to 150GB at SAN level
• diskinfo shows correct 240GB size on primary nodes
• Primary node: Successfully extended using vgmodify, lvextend, fsadm
• Filesystems now 240GB and Oracle running normally

Challenge:

Secondary node has inactive/shared VG - cannot verify LV sizes or filesystem capacity without activating VG. Standard LVM commands (vgdisplay, lvdisplay) fail on inactive VGs.

Current Status:

• Primary node: VG active, all filesystems extended to 240GB
• Secondary node: VG inactive, cannot verify LV configuration
• ServiceGuard package running normally on primary

Questions:

1. How to verify secondary node sees extended LV sizes without activating VG?
2. Does LVM metadata sync automatically to shared storage, or additional steps needed?

Environment Details:

• HP-UX 11.31, ServiceGuard A.11.20
• LVM with shared storage (3PAR)
• VxFS filesystems
• Oracle databases

Looking for ServiceGuard/LVM experts who have handled similar storage expansions in clustered environments.

[mbarnwal]

While the VG is active and filesystems are mounted on primary node, you can still activate the VG on other node in (only) read-only mode.

vgchange -a r <vg name>

Then you can mount the filesystems and check the size. Again, you would be able to mount the filesystems in read-only mode only.

mount -o ro /dev/<vg name>/<lv name> /mnt

Once done with verification, do not forget to unmount the filesystem and deactivate the vg

umount /mnt

vgchange -a n <vg name>

[coke]

thank you. this tells me we can safely activ ate a VG in ro mode, mount the mounts in ro mode

safely means - it wont affect the actual live mount in pri node. I do not know yet if i can try that yet - but thank you.

However I still wonder what is the approach for a mount which is a package - in  a HP Service guard

There might be scenarios where you need to either add a disk or expand a disk to expand FS/mount point size 

you do all the work in primary node to make LVM aware of increased size - but what do we need to do in secondary node

Thank you again,.

[mbarnwal]

If there are any changes to VG on primary node, like addition/removal of PV, you should redistribute the VG to update /etc/lvmtab on other cluster nodes using vgexport/vgimport method.

Refer "Distributing the Volume Group" in SG guide 

https://support.hpe.com/hpesc/public/docDisplay?docId=a00062147en_us&docLocale=en_US

[coke]

Thank you

correct me if I am wrong, pleasse

it looks like a new VG set up

my scenario is the VG already exists in both nodes

in node 1 the disk has been expanded then in OS I made lvm aware of it by vgmodify, lvextend and fsadm

what do i do in node 2 (vg is inactive as it is active in node1)

this VG/mount is part of a cluster pkg

[shiva_jr]

Hi @coke , 
    Yes, I agree with @mbarnwal .   As I remember, cluster node 2 will not update automatically once you change in VG of node 1. 
      Best method is to use the vgexport/vgimport method.  Here are steps-
      a. Use 'vgexport' command to export the vg map file in node 1 which has the modified VG.
      b. Move the file to node 2 using tool such as Winscp.
      c. Use 'vgexport' command to backup the vg map file in node 2.
      d. Use 'vgimport' command to import the VG metadata of node1 to node2.
      e. Run the command 'vgdisplay' or 'bdf' to view the VG details.
For more information, you can refer this post. 
Regards,
Shiva_jr