HACMP Disk Addition

[aixconny]

In a scenario where I have an existing Volume Group which is defined in a 2 node HACMP Cluster's resource group that fails between the two nodes do I *NEED* to run a full HACMP synchronize after I use CSPOC to add the disk(s) to the Volume group?

I have always been extremely careful when adding disks to a HACMP controlled VGs.

My theory is that I do not need to run a sync for the following reason;

The Resource group *only* knows about the VG name and therefore the VGDA of the Volume Group is the real owner of that information i.e what disks are members of the volumegroups etc... so as long as I run the addition through CSPOC both nodes odm's are up to date with disk info.. and as the cluster definition itself doesn't contain specific disk info it should be ok?

I would be grateful if someone could add someone could give me some input just to clarify my theory...

I have always tended to recommend outages so I can do this disk add in the safest way. I know that the cluster services can get flakey at best when multiple syncs are ran with the cluster up...

Thanks..

G

 

[ogniemi]

you don't need to run sync cluster after adding new disk(s) to shared VG in a cluster using C-SPOC.

but of course, before you start adding new disk(s) to the VG you should have already configured new drive(s) with set PVID(s) in all cluster nodes which belongs to "node memebers" list of the resource group having the VG configured in.

 

[khalidaaa]

No need for running sync after adding the disk!

More info:

http://www.tek-tips.com/viewthread.cfm?qid=1440380&page=1

Regards,
Khalid

 

[aixconny]

Thanks Khalid for the link to the previous thread.. sorry for double posting..

Thanks ogniemi!

G

 

[aixconny]

I would really appreciate some help with a few concepts with making changes whilst a cluster is running..

Can I ask you a few more questions from your experience;
I was wondering what **general** rules you guys apply to clustering in relation to changes to a running cluster configuration

I am happy on how to make these changes just wanted to get a feel on how safe they would be whilst the cluster is up and running..

A few tasks that I would really appreciate some clarification on are;

Adding new VG's to the cluster (enhanced concurrent)
Adding Disks to existing VGs (enhanced concurrent) - think we have already covered this one..
Changes to Resource Group definition.
Changes to Topology i.e new IP's, new network cards, new networks etc..
Changes to application server config..

What changes can be performed with little or no risk?

What changes can be performed but are not recommended?

And finally what definitely must not be done with the cluster running?

I am thinking of opening an IBM support call to get the official line but it would be really cool if maybe I could get some of your general experiences as well..

Thanks again

G