OCFS2 Integration with HP Serviceguard for Linux Administrator's Guide, Second Edition, May 2009

slots that are used by the cluster nodes to perform the disk heartbeat. The total number of cluster
nodes which can heartbeat are less than or equal to the value of the -N option.
If, at a later time, you need to increase the number of nodes in the cluster, change the value of
the -N option. However, once set at a higher value, the value of -N cannot be set to a lesser value.
In the example shown above, -N option is set at 4 which creates 4 node slots on the device, /dev/
sdf1. This value allows a minimum of 1 and a maximum of 4 nodes to mount and use the OCFS2
file system on the device, /dev/sdf1. If, at a later time, you need 4 additional nodes to mount
and use the OCFS2 file system configured on the device, /dev/sdf1, then change the value of
the -N option from 4 to 8. With this new value, a minimum of 1 and a maximum of 8 nodes are
concurrently allowed to mount and use the OCFS2 file system on the device.
Controlling OCFS2 Resources using Serviceguard Packages
Serviceguard uses a System Multi-node package to start the OCFS2 stack on all nodes of the
cluster. Once the OCFS2 stack is up and running, Serviceguard mounts the devices that have the
OCFS2 file system on multiple nodes, using the OCFS2 mount point packages. These mount
point packages can each mount multiple OCFS2 devices and can be configured to run on all or
a subset of the total number of nodes in a cluster.
The mount point packages are dependent on the OCFS2 system multi-node package. This
dependency ensures that devices are mounted only after the OCFS2 stack is up and running on
the nodes.
Applications can be packaged in a regular failover or a multi-node package. The application
package must be set up with a dependency on all the OCFS2 mount point packages, which control
the disk resources that the application needs to use.
The subsequent sections describe the following packages:
• System Multi-node Package
• Mount Point Packages
Using the System Multi-node Package
The System Multi-node Package is also referred to as the SG-OCFS2-pkg. This package controls
the OCFS2 cluster stack. It sets the OCFS2 timeout values, creates the cluster configuration and
even starts and stops the cluster. Use the ocfs2cluster command to create the System
Multi-node Package.
For more information on creating a System Multi-node Package using the ocfs2cluster
command, see “Using the ocfs2cluster Command” (page 25).
Using the Mount Point Packages
The OCFS2 Mount Point Packages control disks that contain the cluster file system. These Mount
Point Packages are created with the default name, SG-OCFS2-MP-#. However, you can also
create packages with a different name, such as the disk group package.
These packages mount and un-mount disks that contain the cluster file system on multiple nodes
from where the file system can be accessed simultaneously. You can configure multiple OCFS2
devices in a single mount point package. Configure all the mount point packages to be dependent
on the System Multi-node package. This dependency ensures that the mount operation starts
only after the cluster stack is up and running on the nodes.
Use the ocfs2mntadm command to create, configure, modify, view, start and stop the OCFS2
mount point multi-node packages.
For more information on using the ocfs2mntadm command for creating and modifying a Mount
Point Multi-node package, see “Using the ocfs2mntadm Command” (page 26).
Controlling OCFS2 Resources using Serviceguard Packages 23