Managing HP Serviceguard A.12.00.00 for Linux, June 2014
1. Set the LOAD_BALANCING parameter to ON.
2. Configure one capacity for each node in the cluster. The capacity value can either have a
INFINITE (for all the nodes in the cluster) or finite value.
Package configuration file
In the package configuration file:
1. Set the failover_policy parameter to CONFIGURED_NODE.
2. Set the failback_policy parameter to MANUAL.
3. Configure a weight corresponding to the capacity defined in the cluster configuration file.
4.9.10.8.7 Starting Load Sensitive Package Placement
You can start the package on the most eligible node in one of the following ways:
1. Start the package using the cmrunpkg command:
cmrunpkg –a <pkg_name>...
NOTE: cmrunpkg <pkg_name> always starts the package on the local node only. Hence,
there might be an imbalance in cluster when the load balancing is configured.
2. Apply the package configuration and enable autorun for the packages, when cluster is
down:
cmruncl
3. Enable autorun for a package, so that the package starts on the most eligible node:
cmmodpkg –e <pkg_name>...
NOTE:
• The most eligible node is the node which has the least capacity used (when capacity is
INFINITE) or the node with maximum remaining capacity (when capacity is finite).
• In the cmviewcl -v -f line output capacity:x|percent_used field is always zero
when the corresponding capacity is infinite.
4.9.10.9 Examples
CAPACITY_VALUE parameter set to INFINITE
Scenario 1: Staring a package with cmruncl command
1. Set the LOAD_BALANCING parameter to ON:
Parameters in cluster configuration:
LOAD_BALANCING ON
2. Create a cluster with nodes having CAPACITY_VALUE parameter set to INFINITE:
NODE_NAME test1
CAPACITY_NAME test_capacity
CAPACITY_VALUE INFINITE
3. Repeat step 2 for all the nodes in the cluster.
4. Create packages with weight_name and weight_value in the package configuration file:
weight_name test_capacity
weight_value <value>
For example, consider creating 5 packages with different weight_value as shown:
4.9 Package Configuration Planning 127