最近尝试了一下将Oracle Clusterware 11gR1的OCR和VD放在OCFS2上，这里记录一下OCFS2的安装过程。

首先是确定所使用的Linux的发行版本（distribution），内核版本（version）和风格（flavor），处理器架构：

# cat /etc/redhat-release
Enterprise Linux Enterprise Linux AS release 4 (October Update 5)

# uname -r
2.6.9-67.0.4.0.2.ELsmp

# uname -p
i686

可以看到这里我使用的是Oracle Enterprise Linux 4u5（可以认为是等同于Red Hat Enterprise Linux 4u5），内核版本是2.6.9-67.0.4.0.2，内核风格是smp，处理器架构是i686（也就是x86）。

接下来按照上面的信息去下载相应的OCFS2和OCFS2 Tools（包括ocfs2-tools和ocfs2console）的RPM包。

# wget -c http://oss.oracle.com/projects/ocfs2/dist/files/RedHat/RHEL4/i386/1.2.9-1/2.6.9-67.0.4.EL/ocfs2-2.6.9-67.0.4.ELsmp-1.2.9-1.el4.i686.rpm
# wget -c http://oss.oracle.com/projects/ocfs2-tools/dist/files/RedHat/RHEL4/i386/1.2.7-1/ocfs2-tools-1.2.7-1.el4.i386.rpm
# wget -c http://oss.oracle.com/projects/ocfs2-tools/dist/files/RedHat/RHEL4/i386/1.2.7-1/ocfs2console-1.2.7-1.el4.i386.rpm

删除掉现有的老版本的OCFS2相关RPM包：

# rpm -qa | grep ocfs | xargs rpm -e

依次安装刚刚下载的3个RPM：

# rpm -ivh ocfs2-tools-1.2.7-1.el4.i386.rpm
# rpm -ivh ocfs2console-1.2.7-1.el4.i386.rpm
# rpm -ivh --nodeps ocfs2-2.6.9-67.0.4.ELsmp-1.2.9-1.el4.i686.rpm

注意安装OCFS2 RPM时需要加上“–nodeps”选项。

这里还有一个问题，上面安装的OCFS2 RPM的内核版本（2.6.9-67.0.4）与我所使用的Linux的内核版本（2.6.9-67.0.4.0.2）并不完全一致，所以还需要进行下面的额外工作：

# cp -r /lib/modules/2.6.9-67.0.4.ELsmp/kernel/fs/* /lib/modules/2.6.9-67.0.4.0.2.ELsmp/kernel/fs

至此，OCFS2的安装已经完成。

OCFS2

Open-iSCSI

A discussion on OCFS, OCFS2, ASM, RAW

In 10gR2, we can use:

ocrconfig -export <filename> [-s online]

to export OCR contents to a file.

And use:

ocrconfig -import <filename>

to import OCR contents from a file.

The synchronization of data in multiple caches so that reading a memory location through any cache will return the most recent data written to that location through any other cache. Sometimes called cache consistency.

Cache Fusion

A diskless cache coherency mechanism in Real Application Clusters that provides copies of blocks directly from a holding instance’s memory cache to a requesting instance’s memory cache.

Cluster Manager (CM)

An operating system-dependent component that discovers and tracks the membership state of each node by providing a common view of cluster membership across the cluster. The CM also monitors process health, specifically the health of the database instance.

Enqueue

Shared memory structures that serialize access to database resources. Enqueues are local to one instance if Real Application Clusters is not enabled. When you enable Real Application Clusters, enqueues can be local or global to a database.

Failover

The means of failure recognition and recovery used by Real Application Clusters.

Global Cache Service (GCS)

The process that implements Cache Fusion. It maintains the block mode for blocks in the global role. It is responsible for block transfers between instances. The Global Cache Service employs various background processes such as the Global Cache Service Processes (LMSn) and Global Enqueue Service Daemon (LMD).

Global Enqueue Service (GES)

This service coordinates local and global enqueues.

Global Resource Directory (GRD)

The data structures associated with global resources. It is distributed across all instances in a cluster.

Latch

A simple, low-level serialization mechanism that protect in-memory data structures in the System Global Area (SGA). Latches do not protect datafiles, are automatic, and are held for a very short time in exclusive mode. Because latches are synchronized within a node, they do not facilitate internode synchronization.

Row Cache

The memory that stores recently accessed data for an individual record (or row). Row caches are used so that subsequent requests for the data held in the same row can be processed more quickly.