Managing Serviceguard Nineteenth Edition, Reprinted June 2011

IMPORTANT: It is recommended to increment the parameter values in smaller numbers and
monitor the effect on performance at each level. As soon as the performance starts to decline, stop
increasing and reset it to a lower level. However, you must consider the factors like the number of
CPUs, the amount of available memory, the HP-UX kernel settings, and the number and
characteristics of other packages that will be running on the node.
NOTE: Serviceguard does not support parallel deactivation of disk groups and parallel unmounting
of mount points. These operations are done serially.
Creating Modular Checkpoint and Snapshot Packages for CFS
Mount Point Packages for Storage Checkpoint using Modular Style Packages
The following example illustrates how to create a storage checkpoint:
1. Create a package configuration file for the storage checkpoint:
cmmakepkg -m sg/cfs_all /etc/cmcluster/ckpt1.ascii
Package template is created. This file must be edited before it can be used.
2. Edit the following package parameters in the ckpt1.ascii package configuration file:
package_name ckpt1
node_name node1
node_name node2
ckpt_mount_point /ckpt1
ckpt_name check1
ckpt_cfs_mount_point /mnt2
ckpt_mount_options "node1=cluster node2=cluster"
ckpt_primary_policy node1 node2
NOTE: By default, the package parameters are commented out. You must uncomment them
and specify the values.
ckpt_mount_point name of the mount point for the storage checkpoint
ckpt_name name of the storage checkpoint
ckpt_cfs_mount_point unique name for CFS mount point; supports up to two levels
of nested mount points
ckpt_mount_options the CFS mount options for the storage checkpoint. For detailed
information about the mount options, see the mount_vxfs
(1m) manpage.
ckpt_primary_policy sets the primary migration policy for the mount point. This
option sets up the order in which hosts assume the cluster file
system primary ownership if the primary node fails. If no host
specified in the list is active when the primary node fails, a
node is chosen at random to be the new primary. When the
policy is set, if the first host name specified in the list is not
the current primary, a primary migration is triggered to the
first host in the list. For details, see the fsclustadm (1m)
manpage.
Of the parameters listed above, ckpt_primary_policy can be edited even when the
package is running.
3. Configure the appropriate dependency parameters of the checkpoint:
dependency_name cfspkg1
dependency_condition cfspkg1 = UP
dependency_location same_node
Configuring the Cluster 199