You can install Serviceguard on an HP-UX guest to provide high availability
for the applications running on the guest. In this type of configuration,
the guest is configured as a node in a Serviceguard cluster. Depending on
the configuration of the cluster, the application package can fail over from
one guest to another guest in the same VM Host system, from one guest to another
guest in a VM Host system, or from the guest on a VM Host system to a separate
physical server or nPar. You can even mix and match Serviceguard in Guest
configurations to meet your specific requirements. The following sections
describe the Serviceguard in Guest configurations.
Cluster in a Box |
 |
Figure 10-1 shows the configuration
of an application package that can fail over to another guest on the same
VM Host system.
In this configuration, the primary node and the adoptive node are guests
running on the same VM Host system. This cluster does not provide protection
against Single Point of Failure (SPOF), because both the primary cluster member
and the adoptive cluster member are guests on the same physical machine. However,
this configuration is useful in testing environments.
If you are running more than one guest on the VM Host system, and you
need to share the same storage among the guests, you must change the SHARE
attribute of the shared disk to YES using the hpvmdevmgmt command,
as follows:
# hpvmdevmgmt -m gdev:/dev/rdsk/c6t1d4:attr:SHARE=YES |
For
more information about using the hpvmdevmgmt command, see “Managing the Device Database”.
Virtual/Virtual Cluster |
|
Figure 10-2 shows the configuration
of an application package that can fail over to a guest running on a different
VM Host system.
In this configuration, the Serviceguard nodes are guests running on
separate nPars or HP Integrity servers.
Virtual/Physical Cluster |
|
Figure 10-3 shows the
configuration of an application package that can fail over to a dedicated
HP Integrity server or nPartition.
In this case, the Serviceguard cluster consists of a VM Host system
and a Serviceguard node that is not running Integrity VM. The application
configured as a Serviceguard package can fail over to the physical node. Alternatively,
you can run the application on the physical node and configure the guest on
the VM Host system as the adoptive node.
Configuring Serviceguard in Guests |
|
To configure a Serviceguard cluster that allows an application to fail
over from one guest to another, complete the following procedure
Install Serviceguard on the HP-UX guests that may run the
application.
For the virtual/physical cluster, install Serviceguard on
the physical node.
Ensure that each guest has access to a quorum server or cluster
lock disk.
Use the hpvmstatus command to make sure
the guest is running and to verify the guest name.
Use the cmquerycl command to specify the
nodes to be included in the cluster and to generate a template for the cluster
configuration file. For example, to set up a cluster named gcluster that
includes nodes host1 and host2,
enter the following command:
# cmquerycl -v -C /etc/cmcluster/gcluster.config -n host1 -n host2 -q quorum-server-host |
Include the —q option if a quorum server is used on
the cluster.
Edit the /etc/cmcluster/cluster-name.config file
(where cluster-name is the name of the cluster
specified in the cmquerycl command). For details about
modifying the information in the cluster configuration file, see the Managing
Serviceguard manual.
Use the following command to verify the contents of the file:
# cmcheckconf -k -v -C /etc/cmcluster/gcluster.config |
This command ensures that the cluster is configured properly.
Generate the binary configuration file and distribute it using
the following command:
# cmapplyconf -k -v -C /etc/cmcluster/gcluster.config |
Start the cluster using the following command:
This procedure provides a simple example of creating guest application
packages. For information about how to set up your Serviceguard configuration,
see the Managing Serviceguard manual.
Printable version
