Service Guard A.11.20

[shabash]

i have to configure three node cluster.

Now for node A moutpt failover to node C
Node Node B mountpt failover to Node c
Node C has separate mtpt that will be failover to Node A and Node B.

Now Node A and Node B running 2 database instances and one at a time failover to node C

Also node A mtpts map to node C
Node B mtpts map to Node C
Node C specific mt map to node a and NODe b

******************************


Question is?
As node A and node B has diffrent mtpt and they are not shared between node A and Node b

So in cluster file all vg will be included??

Or

In cluster file Node A vgs include or wht??

what i understand is that for cluster file all mtpts should be shared and for package its a priority setting which enable packge to run at.

Any help???IDea?

[Vivek_Pendse]

Yes, you are correct.
You need to create the package mount points on primary & secondary nodes.
When you will be creating a cluster file, you need to define all the vgs which are contributing in cluster. i.e. cluster vgs. Because, cluster file will be unique distributed across all cluster nodes.
(Or, even you can create a cluster ascii only with lock disk, further, you can add package vgs in respected packages).

Thanks,
Vivek

[N,Vipin]

It is madatory that storage resource must be shared to all nodes participating in the cluster.

[Stephen Doud]

A Serviceguard package will expect the same system resources (LANs, disks, volume groups, mount directories) to exist on any system that runs the package if the package is to start up properly. That means that each 'adoptive' node of the package should be configured similarly.

Page 132 of the A.11.20 'Managing Serviceguard' manual at http://docs.hp.com/en/ha.html -> http://www.hp.com/go/hpux-SG-docs ->
states:
"The volume groups that contain high-availability applications, services, or data
must be on a bus or busses available to the primary node and all adoptive nodes."

This means that if the arrays have the ability to zone disks for visibility to servers, the LUNs that each server should have access to need to be zoned to those servers.
Similarly, if you build LVM volume groups to support the package application on one node, the adoptive nodes must have the LVM structures in place to be able to activate the volume groups when the package is started on that node.

Use the search feature of the forum to locate threads that discuss how to vgimport volume groups using map files, if you need that guidance.
Better still, I recommend that you download the manual and use it to prepare your cluster.