C H A P T E R 11 |
For a description of the options for adding a diskless node or dataless node to a cluster, see the following sections:
Before adding a node to a cluster, ask the following questions:
Is the node of a supported hardware type?
For information about supported hardware types, see the Netra High Availability Suite 3.0 1/08 Release Notes.
How was the cluster originally installed?
The way you add a node to a cluster depends on how the cluster was originally installed. A cluster can be installed by the nhinstall tool or manually. To determine whether a cluster was installed by the nhinstall tool or manually, consult the system operator who installed the cluster.
Is the node already defined in the cluster configuration?
A node that is defined as part of a cluster but is not physically connected to the cluster is called a predefined node. To find out if your cluster contains predefined nodes, see To Determine Whether the Cluster Has Predefined Nodes.
For information about how to add a node to a cluster installed by the nhinstall tool or manually, see the Netra High Availability Suite 3.0 1/08 Foundation Services Manual Installation Guide for the Solaris OS.
To use the CMM API to add or remove a node from a cluster, see the cmm_config_reload3CMM man page.
To determine whether a cluster has predefined nodes, perform the following procedure.
Determine how many nodes are up and running in the cluster:
# nhcmmstat –c all |
This command returns information about the nodes that are up and running in the cluster, but does not return information about nodes that are physically present but down.
Determine how many nodes are defined in the cluster configuration.
Look at the cluster_nodes_table file.
The file contains a line for each node in the cluster definition. Count the number of nodes configured in this file. On Solaris systems, the file is located at /etc/opt/SUNWcgha/cluster_nodes_table. On Linux systems, the file is located at /etc/opt/sun/nhas/cluster_nodes_table.
If there is an /export/root directory on the master node, examine its contents.
There is one /diskless-node-name directory for each diskless node in the cluster.
Count the number of nodes physically present in the cluster.
Use the output of the nhcmmstat command and your knowledge of any nodes that are physically present but down. If a cluster has predefined nodes that are not diskless nodes, they are dataless nodes.
Copyright © 2008, Sun Microsystems, Inc. All rights reserved.