HP-UX LVM Storage Extension

[coke]

Need to extend  filesystems from 180GB to 240GB each on HP-UX 11.31 with ServiceGuard cluster, but hitting VG physical extent limitations.

• HP-UX 11.31 with ServiceGuard cluster
• VxFS filesystems
• LVM with 8MB PE size
• Storage team expanded LUNs from 180GB to 240GB at SAN level

vgmodify -E -a vgname

fails with: "Max_PE_per_PV for the volume group (23039) too small for this PV (30719)"

Current VG settings limit expansion:

• Max PE per PV: 23,039 (allows ~180GB per PV)
• Required: 30,720 PEs (for 240GB per PV)
• Max PV: 16, VG Max Extents: 368,624

How to increase Max PE per PV limit in HP-UX LVM without VG recreation or downtime? Standard pvresize/lvextend won't work due to VG parameter constraints.

• ServiceGuard cluster environment

# vgmodify -v -r -a -E VGName
Volume Group configuration for /dev/vVGName has been saved in /etc/lvmconf/VGName.conf
/dev/rcdisk/disk226 Warning: Max_PE_per_PV for the volume group (23039) too small for this PV (30719).
Using only 23039 PEs from this physical volume.


Using only 23039 PEs from this physical volume.
"/dev/rcdisk/disk226" size is limited by the maximum extents per PV to 188743679 KB.

An update to the Volume Group is NOT required
Review complete. Volume group not modified

#

I fixed it by reducing max PV form 16 to 10

then attempt vgmodify again

"/dev/rcdisk/disk226" size changed from 188743680 to 251658240kb

lvextend and then fsadm

The filesystes or mounts constitute a SG package which is active on node 1 now.

failover node is node 2

how do I / do i need to make that node 2 aware of this increased disk size (or increased FS or mount size)

if you have any questions, need more information let me know pls

[shiva_jr]

Hi @coke ,
    The Strict allocation policy might be stopping you to extend. I believe, you have to move the Physical extents using pvmove command and make the PEs available continuously.  Please refer this video.
Regards,
Shiva_jr

[coke]

@shiva_jr, Thank you

I got around that by running vgmodify.. I reduced max pv parameter value from 16 to 10 amd it allowed me to expand vg

all this was done in pri node where the vg is active and cluster package is running on node1

The filesystes or mounts constitute a SG package which is active on node 1 now.

failover node is node 2

how do I / do i need to make that node 2 aware of this increased disk size (or increased FS or mount size)

[coke]

in node 1

VG is active and has three LVs / mount points associated with it.

each LV has one disk

three disks' size was increased from storage side.

max PEs that can be allocated was 100 let us say. and all of the (allocatable) PEs were allocated  already with existing PV size

vgmodify (with decreased 'max pv' and increased PE) lvextend and fsadm - completed successfully

so now with the modified VG I managed to increase the LV / mount size in node1

This modified VG is part. of a package  (HP service guard)

this VG, this pkg is active on node1

--

Now with VG modified in node1 

with VG inactive in node 2 - how do me make this node2's VG updated with the changes we did

vgexport to create a mapfile in node1 - then copy over the mapfile to node2.

what would be the exact command syntax in node1 and node2

more importantly - without activating VG in node2 - how do we verify the VG is updated in node2