Managing Serviceguard Nineteenth Edition, Reprinted June 2011

1. Identify the nodes for which you want to create cluster-wide device files; this is known as the
cDSF group.
This should be all the nodes in the cluster (or prospective cluster).
IMPORTANT: Remember that:
• You cannot create a cDSF group that crosses cluster boundaries; that is, the group must
consist of the nodes of a single cluster.
• cDSFs use agile addressing; see “About Device File Names (Device Special Files)”
(page 77) for information about agile addressing.
2. If the cluster does not yet exist, set up root access among the prospective nodes:
a. If you have not already done so, set up ssh public/private key pairs on each node. This
will allow the necessary commands to operate on all the prospective nodes before a
cluster is formed.
The simplest way to do this is via the DSAU csshsetup command; for example, if you
are setting up a two-node cluster with nodes node1 and node2, and you are logged in
on node1:
csshsetup -r node2
For a large number of nodes, you might want to enter the node names into a file and use
the -f option to get csshsetup to read the names from the file; for example, if you
have stored the names in the file /etc/cmcluster/sshhosts:
csshsetup -r -f /etc/cmcluster/sshhosts
For more information about setting up ssh keys, see the HP-UX Secure Shell Getting
Started Guide at http://www.hp.com/go/hpux-core-docs.
b. Configure root access to each prospective node, using the hostname portion (only) of the
fully-qualified domain name:
cmpreparecl -n <node_name> -n <node_name> ...
For example, for a cluster that will consist four nodes, node1, node2, node3, and
node4:
cmpreparecl -n node1 -n node2 -n node3 -n node4
NOTE: Serviceguard must be installed on all of the nodes listed, and you must be logged
in as superuser on one of these nodes to run the command.
3. Create the cDSFs.
NOTE: cDSFs apply only to shared storage; they will not be generated for local storage,
such as root, boot, and swap devices.
• If the cluster does not exist yet, specify the name of each prospective node, for example:
cmsetdsfgroup -n node1 -n node2 -n node3 -n node4
• If the cluster does exist, you can simply run:
cmsetdsfgroup -c
NOTE: You must be logged in as superuser on one of the cluster nodes. You do not
need to provide the cluster name.
The cDSFs created by cmsetdsfgroup reside in /dev/cdisk for block device files and /dev/
rcdisk for character devicefiles. You should use these new device files exclusively when you
configure the cluster lock (if any) and package storage; see “Specifying a Lock Disk” (page 179),
Preparing Your Systems 151