This is an excerpt from the bestselling book
Oracle Grid & Real Application Clusters. To get immediate
access to the code depot of working RAC scripts, buy it
directly from the publisher and save more than 30%.
PSFS is a general-purpose cluster file system.
It is an enterprise-grade symmetric file system. The general
* PSFS allows Oracle Managed Files (OMF) to be
used in an Oracle RAC cluster. Oracle OMF provides the ability to
manage files and file systems that reduces the complexity of
creating, configuring, and managing the storage for an Oracle
database. Oracle9i RAC with a shared raw device cannot use OMF
* All matrix server members have access to the
file system and can perform concurrent read/write operations.
* Supports Direct I/O facility, which increases
the I/O performance by avoiding the file system buffering
* The Oracle Disk Manager (ODM). ODM is an API
specification for I/O designed by Oracle. PolyServe?s ODM offering,
MxODM, is a plug-in module for PolyServe?s cluster file system.
MxODM comes bundled with either the Oracle RAC Option or Oracle
Option for Matrix Server.
* Device Naming and Utilities ? Cluster-wide
device naming and file system utilities simplify administration and
maintenance of the cluster. MxS provides a pseudo device layer that
ensures consistent device naming across the cluster, and provides
various tools for file system maintenance, including fsck, file
system suspend, and resume and file system growth.
* PSFS also supports context dependant symbolic
links that allow administrators to configure node-specific files.
* MxS supports all Oracle data files and logs
and further supports advanced Oracle9i functionality, including
external tables, export/import from text files, and archived redo
* MxS also supports a shared Oracle Home, which
simplifies installation, configuration, and maintenance.
* It is a standard Linux File system that
supports file system operations such as mkfs, mount and unmount and
byte-range locking with complete cache coherency.
PolyServe MxS uses a registry to control server
access to the PSFS file system on the SAN disks imported into the
matrix. It supports journaling and online crash recovery.
PSFS supports context dependent symbolic links
(CDSL). CDSL are used to manage node-specific files and directories.
CDSL(s) enable database administrators to easily create an Oracle
Home directory that can be shared by all nodes in the matrix.
CDSL contains a keyword that identifies a
particular location. When the operating system resolves the symbolic
link, it translates the keyword into the appropriate pathname. For
example, a CDSL could contain the keyword HOSTNAME, which resolves
the hostname of the server where it has been logged in.
Configure and Create PSFS
The PSFS file system is created on a storage
extent located on imported disks. Storage extents are disk
partitions. The maximum file system is 1 TB. PSFS uses 4KB as the
A file system can be created using either the
management console or the command line. Then, the file system can be
mounted on all the nodes or servers in the matrix. Since the disk
partition is the unit on which the file system is created, plan the
size of the partition carefully.
To create a file system using the CLI, use the
# mx fs create [--size <kb>] <filesystem> <StorageExtent>
# mkpsfs <device> [<size_in_blks>]
Where the global device name is specified as
/dev/psd/psdXXXpYY where XXX is the disk number and YY is the
partition number. Then, mount the PSFS file system on any server in
the matrix. Use the Polyserve MxS command:
# mx fs mount [--persist] [--activate]
[--options] <path> <filesystem>