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

1 Introduction
This chapter elaborates on the features of Oracle Cluster File System (OCFS2) and the benefits
of integrating it with HP Serviceguard for Linux.
This chapter addresses the following topics:
• “Overview of OCFS2” (page 11)
• “Overview of HP Serviceguard for Linux” (page 12)
• “Benefits of OCFS2 in an HP Serviceguard for Linux Environment” (page 13)
Overview of OCFS2
Oracle Cluster File System 2 (OCFS2) is an independent cluster file system. With the built-in
kernel-level membership service and heartbeat functions, OCFS2 is capable of running as a
cluster. By default, OCFS2 uses the disk-based heartbeat. Nodes on which OCFS2 volumes are
mounted regularly write into a metadata heartbeat area in the file system. By doing so, nodes
ensure that they are recognized as live nodes.
When configured, the cluster manager, O2CB, manages the cluster membership. The O2CB
cluster manager service runs on every node in the cluster. It polls the nodes regularly to ensure
that all nodes in the cluster are alive. Communication between O2CB and the nodes is done over
a private network.
The subsequent section elaborates on the features of OCFS2.
Features of OCFS2
OCFS2 is a file system that is integrated with Linux and has in-built kernel-level membership.
Following are the features of OCFS2:
• O2CB Configurable Parameters
The cluster manager, O2CB, has a number of parameters that can be configured to poll the
nodes to ensure that every node in the cluster is alive.
• OCFS2 Quorum Capability and Fencing
In an OCFS2 cluster, the quorum decision is made by a single node by determining the
number of nodes that are alive. This is determined by the number of nodes that respond to
the keep-alive responses and the number of nodes that are reachable through the network.
A node with OCFS2 mounted fences itself when it is removed from the cluster membership
in a degraded cluster. A node fences itself so that other nodes are not obstructed from
accessing its resources.
• Network Based, and Pluggable Distributed Lock Manager (DLM)
This feature tracks all the locks in the cluster. It also tracks information such as the owner
of the locks and the lock status.
• Linux Kernel Journal Block Device subsystem
This feature improves journaling and recovery operations.
• Cross-node File Data Consistency
• Metadata Caching and Journaling
Overview of OCFS2 11