topic Re: controlscript and running cluster in Operating System - HP-UX

controlscript and running cluster

Inesa Clinko — Tue, 20 Sep 2005 09:49:23 GMT

I need to add new VG,LV and FS into
control script of the package:

mypackage.cntl

Need I stop the cluster to apply these changes?

Re: controlscript and running cluster

A. Clay Stephenson — Tue, 20 Sep 2005 09:57:02 GMT

The cluster can remain up while doing any of these changes. If this is an existing package, the package must be halted.

Re: controlscript and running cluster

Uday_S_Ankolekar — Tue, 20 Sep 2005 10:01:35 GMT

No need to halt the cluster but if your adding new VG to existing package then package needs to be halted.
You can then modify control scripts apply changes and start package

-USA..

Re: controlscript and running cluster

Uday_S_Ankolekar — Tue, 20 Sep 2005 10:02:53 GMT

Here is more
http://www1.itrc.hp.com/service/cki/docDisplay.do?docLocale=en_US&docId=200000079973202

Re: controlscript and running cluster

baiju_3 — Fri, 23 Sep 2005 14:38:06 GMT

Hi ,

No need of halting the cluster packages , you can do the modifications online .

For time being you can activate the vg manually and mount file systems after updating the control file .

Next time when you bounce the package every thing will come up normal .

Thanks,
BL.

Re: controlscript and running cluster

Stephen Doud — Mon, 26 Sep 2005 07:05:03 GMT

You need not stop the package.

Create the VG and logical volumes before proceeding. Verify the lvol's will mount, then umount them and deactivate the VG.

1) MAKE CERTAIN that you add the new VG to the cluster configuration ASCII file. Anytime a cmapplyconf is performed on this file, any VG not represented will be de-clustered!!

2) Manually cluster the VG. On a node where the the VG is active, and Serviceguard is running, perform "vgchange -c y ".

3) Create a map file of the VG:
# vgexport -pvs -m /etc/lvmconf/map.vgNAME /dev/vgNAME

4) Copy map.vgNAME to the other node(s) in the cluster.

5) VGimport the new VG into the other node(s):
OTHERNODE:
# mkdir /dev/vgNAME
# mknod /dev/vgNAME/group c 64 0xNN0000 (where NN = a unique minor number)
# vgimport -vs -m /etc/lvmconf/map.vgNAME /dev/vgNAME

6) ON the original node, update the package control script with references to the new VG and it's logical volumes and mount directories (DON'T FORGET to increment the index numbers within the "[ ]")

8) Copy the script to the other server.

9) Activate the VG, mount it's logical volumes.

10) Test package shutdown, startup and failover to the other node.