Sun Microsystems
Products & Services
 
Support & Training
 
 

Previous Previous     Contents     Index     Next Next

The following program listing is a node group and node definition for a diskless node group for a four-node cluster:
# diskless group and related nodes definitions
ELEMENT nodeGroup diskless type DISKLESS os SOLARIS arch SPARC
        INCLUDE node peerNode3-4N
                node peerNode4-4N
        RUN service NHAS_DISKLESS
            service MAC_ADDR_POLICY

#
# diskless nodes definitions
ELEMENT node peerNode3-4N
        USE board CT@peerNode3
            
ELEMENT node peerNode4-4N
        USE board CT@peerNode4

The following program listing is a node group and node definition for a dataless node group for a twelve-node cluster:

# Node Groups definitions
#
# NMEN group and related nodes definitions
ELEMENT nodeGroup dataless_N120 type DATALESS os SOLARIS arch SPARC
        INCLUDE node peerNode3-12N
                node peerNode4-12N
                node peerNode5-12N
                node peerNode6-12N
        RUN service NHAS_DATALESS

# Nodes definitions
ELEMENT node peerNode3-12N
        USE board N120@peerNode3 
            disk disk1@peerNode3
#
ELEMENT node peerNode4-12N
        USE board N120@peerNode4 
            disk disk1@peerNode4 
#
ELEMENT node peerNode5-12N
        USE board N120@peerNode5 
            disk disk1@peerNode5 
#
ELEMENT node peerNode6-12N
        USE board N120@peerNode6 
            disk disk1@peerNode6 

Choosing Services for Each Node Group

The services to be run on each node group are specified in the node group definition section of the cluster.conf file.

For a complete list of the core services, the optional services, and the diskless node boot policies, see Planning Services for Your Cluster.

ProcedureTo Choose Services for Each Node Group

  1. Define the core services for each node group.

    These services are mandatory. These services are defined by predefined core services lists.

    The core service list that you assign to each node group depends on the role of the node group. That is, NHAS_MASTER_ELIGIBLE for a master-eligible node group, NHAS_DISKLESS for a diskless node group, or NHAS_DATALESS for a dataless node group.

  2. Define any optional services to run on each node group.

    This step is optional.

    The optional services are as follows:

    • NSM for the Node State Manager. This services can only be assigned to master-eligible node groups.

    • The watchdog timer service can be assigned to any type of node group. The watchdog timer services per node group are: WDT_MASTER_ELIGIBLE, WDT_DISKLESS, and WDT_DATALESS

  3. Specify the boot policy and Reliable Boot Service for each diskless node group.

    • The boot policy must be defined for each diskless node group. The boot policy can be STATIC_CLIENT_ID_POLICY or MAC_ADDR_POLICY.

    • Ensure that the RBS has been assigned in the relevant master-eligible group service list. This service is required for master-eligible node groups containing diskless node groups. In addition, this service can only be assigned to master-eligible node groups with diskless node groups.

In the following example of a service definition, the core service list for the master-eligible node is NHAS_MASTER_ELIGIBLE, and the optional service is NSM:

RUN service NHAS_MASTER_ELIGIBLE
service NSM 

For information on the full range of services and service lists available for the node groups, see the cluster.conf(4) man page.

Defining the Boot Policy for a Diskless Node Group

You must specify one of the following boot policies for each diskless node group:

If your diskless nodes are CompactPCI boards you can choose between MAC_ADDR_POLICY or STATIC_CLIENT_ID_POLICY.

If your diskless nodes are other types of nodes, you must choose MAC_ADDR_POLICY.

In the node group definition section of the cluster.conf file, use the following syntax to specify the boot policy:

RUN service core_service_ list
    service boot_policy

For example:

RUN service NHAS_DISKLESS
    service MAC_ADDR_POLICY

For more information, see the cluster.conf(4) man page.

Previous Previous     Contents     Index     Next Next