Adding new file system in existing running SG Cluster

[unix_hp_1]

Dear Experts,

i have 2 node active-passive service guard cluster environment running on HP-UX 11iv3.

our storage team has allocated space on my server which is shared on all cluster node, and i have also created "newvg1" of 100gb on running node.
Now i want to add this new vg and its Lv in my cluster so in the fail over situation it will be automatically available on failover node.

please Guide me to do this configuration in step by step with command detail.

Thanking you all in advance.

[Stephen Doud · Accepted solution]

One NODE1 do:
Create the logical volumes and file system(s) and mount directory for newvg1.
Ceate a LVM map file of newvg1:
# vgexport -pvs -m newvg1.map /dev/newvg1
Copy newvg1.map to NODE2

On NODE2, do:
mkdir /dev/newvg1
mknod /dev/newvg1/group c 64 0xNN0000 (legacy style VGs)
vgimport -vs -m newvg1.map /dev/newvg1

On one server update the cluster ASCII file with the new VOLUME_GROUP name, and cmapplyconf the cluster ASCII file. (It may be necessary to deactivate the VG befoer this step)

Edit the package configuration file (modular package) or package control script (legacy package) with references to the VG name, Lvol name(s) and file system mount directory and mount options.

cmappyconf the package configuration file (modular) or scp the package control script (legacy) to the other node.

Test package startup/shutdown and failover

[P Arumugavel]

Hi,

When you perform any modification in LVM, it has to get updated in others noded in cluster. So it is necessary to do vgimport on the adoptive nodes.

Now you had created a volum group named "newvg1". And you will create needed logical volumes and filesystems for the package.
Now perform vgexport to export the vg information from the fist node.
#vgexport -v -p -s -m /tmp/newvg1.map /dev/newvg1

Now on the secondary node,

#rcp primarynode:/tmp/newvg1.map /tmp
#mkdir /dev/vgnew1
#mknod /dev/vgnew1/group c 64 0xYY0000
Note: minor number 0xYY0000 should be same as used in primary node
#vgimport -v -s -m /tmp/vgnew1.map /dev/vgnew1

Now create mount points for needed filesystems.

Then go to primary node, and mark the vgnew1 as cluster awared.
#vgchange -c y vgnew1
Now unmount all the filesystems in this VG and deactivate and halt the package that will use this volume group.
#vgchange -a n vgnew1
#cmhaltpkg

Modify the cluster conf file and package control scripts to insure this new vg is activated and all filesystems are mounted when the package starts.

#cmrunpkg -v pkgname
#cmmodpkg -v -e pkgname

Test to ensure failing over the package onto the adoptive node should verify the new volume group and filesystems are availble on the secondary node.

Rgds...

[sarfaraj ahmad]

Hi,
Please go through the below thread which will be helpful for you in this regard,

http://h30499.www3.hp.com/t5/Serviceguard/create-new-vg-under-SG-cluster/m-p/5267896#M25335

Thanks

[unix_hp_1]

Hi All,

Thanks for your expert reply.

means there is downtime required to adding the new file system in cluster.

[sarfaraj ahmad]

Yes. you should halt the package then update the cluster ASCII file and package configuration file to make the entry for new VG. so this filesystem will be activated automatically in the failover scenario.

[Deeos]

Hi ,


please find the attached procedure


Regards
Deeos