VXFS 5.0

Veritas File System
Administrator's Guide
Solaris
N18516F
Veritas File System Administrator's Guide
The software described in this book is furnished under a license agreement and may be used
only in accordance with the terms of the agreement.
Veritas File System 5.0
PN: N18516F
Legal Notice
Copyright © 2006 Symantec Corporation.
All rights reserved.
Federal acquisitions: Commercial Software - Government Users Subject to Standard License

Terms and Conditions.
Symantec, the Symantec Logo, and Storage Foundation are trademarks or registered
trademarks of Symantec Corporation or its affiliates in the U.S. and other countries. Other
names may be trademarks of their respective owners.
The product described in this document is distributed under licenses restricting its use,
copying, distribution, and decompilation/reverse engineering. No part of this document
may be reproduced in any form by any means without prior written authorization of
Symantec Corporation and its licensors, if any.
THE DOCUMENTATION IS PROVIDED "AS IS" AND ALL EXPRESS OR IMPLIED CONDITIONS,
REPRESENTATIONS AND WARRANTIES, INCLUDING ANY IMPLIED WARRANTY OF
MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE OR NON-INFRINGEMENT,
ARE DISCLAIMED, EXCEPT TO THE EXTENT THAT SUCH DISCLAIMERS ARE HELD TO
BE LEGALLY INVALID. SYMANTEC CORPORATION SHALL NOT BE LIABLE FOR INCIDENTAL
OR CONSEQUENTIAL DAMAGES IN CONNECTION WITH THE FURNISHING PERFORMANCE,
OR USE OF THIS DOCUMENTATION. THE INFORMATION CONTAINED IN THIS
DOCUMENTATION IS SUBJECT TO CHANGE WITHOUT NOTICE.
The Licensed Software and Documentation are deemed to be "commercial computer software"
and "commercial computer software documentation" as defined in FAR Sections 12.212 and
DFARS Section 227.7202.
Symantec Corporation 20330 Stevens Creek Blvd. Cupertino, CA 95014 USA
http://www.symantec.com
Printed in the United States of America.
10 9 8 7 6 5 4 3 2 1
Contents
Chapter 1 Introducing Veritas File System

About Veritas File System .............................................................. 11
Logging ................................................................................ 12
Extents ................................................................................ 12
File system disk layouts ........................................................... 12
Veritas File System features ........................................................... 12
Extent-based allocation ........................................................... 14
Extent attributes .................................................................... 16
Fast file system recovery .......................................................... 16
Extended mount options .......................................................... 17
Enhanced data integrity modes ................................................. 18
Enhanced performance mode ................................................... 19
Modes of temporary file systems ............................................... 19
Improved synchronous writes ................................................... 19
Support for large files ............................................................. 20
Access Control Lists ................................................................ 20
Storage Checkpoints ............................................................... 20
Online backup ....................................................................... 20
Quotas ................................................................................. 21
Support for databases ............................................................. 21
Cluster file systems ................................................................ 22
Cross-platform data sharing ..................................................... 22
File Change Log ..................................................................... 22
Multi-volume support ............................................................. 23
Dynamic Storage Tiering ......................................................... 23
Veritas File System performance enhancements ................................. 23
About enhanced I/O performance .............................................. 24
Using Veritas File System .............................................................. 25
Veritas Enterprise Administrator Graphical User Interface ............. 25
Online system administration ................................................... 26
Application program interface .................................................. 27
4 Contents
Chapter 2 VxFS performance: creating, mounting, and tuning

File Systems
mkfs command options ................................................................. 29
Block size ............................................................................. 29
Intent log size ........................................................................ 30
Choosing mount command options .................................................. 30
The log mode ......................................................................... 31
The delaylog mode ................................................................. 32
The tmplog mode ................................................................... 32
The logiosize mode ................................................................. 33
The nodatainlog mode ............................................................. 33
The blkclear mode .................................................................. 33
The mincache mode ................................................................ 34
The convosync mode ............................................................... 35
The ioerror mode ................................................................... 36
The largefiles|nolargefiles option .............................................. 37
The cio option ....................................................................... 39
Combining mount command options ......................................... 39
Using kernel tunables ................................................................... 40
Tuning inode table size ........................................................... 40
vx_maxlink ........................................................................... 41
Veritas Volume Manager maximum I/O size ................................ 41
Monitoring free space ................................................................... 42
Monitoring fragmentation ....................................................... 42
Tuning I/O .................................................................................. 43
Tuning VxFS I/O parameters .................................................... 44
Tunable I/O parameters ........................................................... 45
File system tuning guidelines ................................................... 52
Chapter 3 Extent attributes

About extent attributes ................................................................. 55
Reservation: preallocating space to a file ..................................... 56
Fixed extent size .................................................................... 56
Other controls ....................................................................... 57
Commands related to extent attributes ............................................ 58
Failure to preserve extent attributes ........................................... 59
Chapter 4 VxFS I/O Overview

About VxFS I/O ............................................................................ 61
Buffered and Direct I/O ................................................................. 62
Direct I/O ............................................................................. 62
Contents 5
Unbuffered I/O ...................................................................... 63

Data synchronous I/O ............................................................. 63
Cache advisories .......................................................................... 64
Freezing and thawing a file system .................................................. 65
Getting the I/O size ...................................................................... 65
Getting the I/O size ....................................................................... 65
Getting the I/O size ....................................................................... 65
Getting the I/O size ...................................................................... 66
Chapter 5 Storage Checkpoints

About Storage Checkpoints ............................................................ 67
How Storage Checkpoints differ from snapshots ........................... 68
How a Storage Checkpoint works ..................................................... 69
Copy-on-write ....................................................................... 71
Types of Storage Checkpoints ......................................................... 72
Data Storage Checkpoints ........................................................ 72
nodata Storage Checkpoints ..................................................... 73
Removable Storage Checkpoints ................................................ 74
Non-mountable Storage Checkpoints ......................................... 74
Storage Checkpoint administration .................................................. 74
Creating a Storage Checkpoint .................................................. 75
Removing a Storage Checkpoint ................................................ 76
Accessing a Storage Checkpoint ................................................ 77
Converting a data Storage Checkpoint to a nodata Storage
Checkpoint ..................................................................... 78
Space management considerations .................................................. 86
Restoring a file system from a Storage Checkpoint .............................. 87
Restoring a file from a Storage Checkpoint .................................. 87
Storage Checkpoint quotas ............................................................. 92
Chapter 6 Online backup using file system snapshots

About snapshot file systems ........................................................... 93
Snapshot file system backups ......................................................... 94
Creating a snapshot file system ....................................................... 95
Backup examples .......................................................................... 96
Snapshot file system performance ................................................... 97
Differences between snapshots and Storage Checkpoints ..................... 98
About snapshot file system disk structure ......................................... 98
How a snapshot file system works ................................................... 99
6 Contents
Chapter 7 Quotas
About quota limits ...................................................................... 101
About quota files on &ProductNameLong; ....................................... 102
About quota commands ............................................................... 103
About quota checking with Veritas File System ................................. 104
................................................................................................ 104
Using quotas ............................................................................. 104
Turning on quotas ................................................................ 104
Turning on quotas at mount time ............................................ 105
Editing user and group and group and group and group
quotas .......................................................................... 106
Modifying time limits ............................................................ 107
Viewing disk quotas and usage ................................................ 107
Displaying blocks owned by users or groups .............................. 108
Turning off quotas ................................................................ 108
Chapter 8 File Change Log

About the File Change Log file ....................................................... 109
File Change Log administrative interface ......................................... 110
File Change Log programmatic interface ......................................... 112
Reverse path name lookup ............................................................ 114
Chapter 9 Multi-volume file systems

About multi-volume support ......................................................... 118
About volume types .................................................................... 118
Features implemented using multi-volume support ........................... 118
Volume availability ............................................................... 119
About volume sets ...................................................................... 120
Creating and managing volume sets ......................................... 120
Creating multi-volume file systems ................................................ 121
Example of creating a multi-volume file system .......................... 122
Converting a single volume file system to a multi-volume file
system ................................................................................ 123
Removing a volume from a multi-volume file system ......................... 124
Forcibly removing a volume .................................................... 124
About allocation policies .............................................................. 125
Assigning allocation policies ......................................................... 125
Querying allocation policies ......................................................... 126
Assigning pattern tables to directories ............................................ 127
Assigning pattern tables to file systems .......................................... 127
Allocating data ........................................................................... 128
Contents 7
Volume encapsulation ................................................................. 129

Encapsulating a volume ......................................................... 129
Deencapsulating a volume ...................................................... 130
Reporting file extents .................................................................. 131
Examples of reporting file extents ........................................... 131
Load balancing ........................................................................... 132
Defining and assigning a load balancing allocation policy ............ 133
Rebalancing extents .............................................................. 133
Converting a multi-volume file system to a single volume file
system ................................................................................ 134
Converting to a single volume file system .................................. 134
Chapter 10 Dynamic Storage Tiering

About Dynamic Storage Tiering ..................................................... 137
Placement classes ....................................................................... 139
Tagging volumes as placement classes ...................................... 140
Listing placement classes ....................................................... 140
Administering placement policies .................................................. 140
Assigning a placement policy .................................................. 141
Unassigning a placement policy .............................................. 141
Analyzing the space impact of enforcing a placement policy ......... 141
Querying which files will be affected by enforcing a placement
policy ........................................................................... 142
Enforcing a placement policy .................................................. 142
Validating a placement policy ................................................. 143
File placement policy grammar ..................................................... 143
File placement policy rules ........................................................... 156
SELECT statement ................................................................ 156
CREATE statement ................................................................ 160
RELOCATE statement ............................................................ 162
DELETE statement ................................................................ 173
Calculating I/O temperature and access temperature ......................... 174
Multiple criteria in file placement policy rule statements ................... 178
Multiple file selection criteria in SELECT statement clauses ......... 178
Multiple placement classes in <ON> clauses of CREATE
statements and in <TO> clauses of RELOCATE statements
.................................................................................... 180
Multiple placement classes in <FROM> clauses of RELOCATE
and DELETE statements .................................................. 181
Multiple conditions in <WHEN> clauses of RELOCATE and
DELETE statements ........................................................ 181
File placement policy rule and statement ordering ............................ 181
8 Contents
File placement policies and extending files ...................................... 184
Chapter 11 Quick I/O for Databases

About Quick I/O ......................................................................... 185
About Quick I/O functionality and performance ............................... 186
About asynchronous I/O kernel support .................................... 186
About direct I/O support ........................................................ 186
About Kernel write locks avoidance .......................................... 186
About double buffering avoidance ............................................ 187
About using Veritas File System files as raw character devices ............. 187
About the Quick I/O naming convention ................................... 187
About use restrictions ........................................................... 188
About creating a Quick I/O file using qiomkfile ................................ 188
Creating a Quick I/O file using qiomkfile ................................... 189
Accessing regular VxFS files through symbolic links ......................... 190
About absolute and relative path names .................................... 190
Preallocating files using the setext command ............................. 191
Using Quick I/O with Oracle databases ........................................... 192
Using Quick I/O with Sybase databases ........................................... 192
Enabling and disabling Quick I/O ................................................... 196
About Cached Quick I/O for databases ............................................ 197
Enabling Cached Quick I/O ..................................................... 197
About Quick I/O statistics ............................................................ 199
Increasing database performance using Quick I/O ............................. 199
Appendix A Quick Reference

Command summary .................................................................... 201
Online manual pages ................................................................... 205
Creating a VxFS file system .......................................................... 211
Example of creating a file system ............................................. 211
Converting a file system to VxFS .................................................... 212
Example of converting a file system ......................................... 212
Mounting a file system ................................................................ 213
Mount options ..................................................................... 214
Example of mounting a file system ........................................... 215
Editing the vfstab file ............................................................ 215
Unmounting a file system ............................................................ 216
Example of unmounting a file system ....................................... 216
Displaying information on mounted file systems .............................. 216
Example of displaying information on mounted file systems ......... 217
Contents 9
Identifying file system types ......................................................... 217

Example of determining a file system's type .............................. 218
Identifying file system types ......................................................... 218
Example of determining a file system's type .............................. 219
Resizing a file system .................................................................. 220
Extending a file system using fsadm ......................................... 220
Shrinking a file system .......................................................... 221
Reorganizing a file system ...................................................... 222
Backing up and restoring a file system ............................................ 223
Creating and mounting a snapshot file system ........................... 223
Backing up a file system ......................................................... 224
Restoring a file system ........................................................... 225
Using quotas ............................................................................. 225
Turning on quotas ................................................................ 225
Setting up user quotas ........................................................... 226
Viewing quotas .................................................................... 227
Turning off quotas ................................................................ 227
Appendix B Diagnostic messages

File system response to problems ................................................... 229
Recovering a disabled file system ............................................. 230
About kernel messages ................................................................ 230
About global message IDs ....................................................... 230
Kernel messages ......................................................................... 231
About unique message identifiers .................................................. 272
Unique message identifiers .......................................................... 272
Appendix C Disk layout

About disk layouts ...................................................................... 277
About disk space allocation .......................................................... 278
VxFS Version 4 disk layout ........................................................... 279
Using UNIX Commands on File Systems Larger than One TB .............. 283
Glossary
Index
10 Contents
Chapter 1
Introducing Veritas File
System
This chapter includes the following topics:
■ About Veritas File System
■ Veritas File System features
■ Veritas File System performance enhancements
■ Using Veritas File System
About Veritas File System

A file system is simply a method for storing and organizing computer files and
the data they contain to make it easy to find and access them. More formally, a
file system is a set of abstract data types (such as metadata) that are implemented
for the storage, hierarchical organization, manipulation, navigation, access, and
retrieval of data.
Veritas File System (VxFS) was the first commercial journaling file system. With
journaling, metadata changes are first written to a log (or journal) then to disk.
Since changes do not need to be to be written in multiple places, throughput is
much faster as the metadata is written asynchronously.
VxFS is also an extent-based, intent logging file system. VxFS is designed for use
in operating environments that require high performance and availability and
deal with large amounts of data.
VxFS major components include:
■ Logging
■ Extents
12 Introducing Veritas File System
Veritas File System features
■ File system disk layouts
Logging
A key aspect of any file system is how to recover if a system crash occurs. Earlier
methods required a time-consuming scan of the entire file system. A better solution
is the method logging (or journaling) the metadata of files.
VxFS logs new attribute information into a reserved area of the file system,
whenever file system changes occur. The file system writes the actual data to disk
only after the write of the metadata to the log is complete. If and when a system
crash occurs, the system recovery code analyzes the metadata log and try to clean
up only those files. Without logging, a file system check (fsck) must look at all of
the metadata.
Intent logging minimizes system downtime after abnormal shutdowns by logging
file system transactions. When the system is halted unexpectedly, this log can be
replayed and outstanding transactions completed. The check and repair time for
file systems can be reduced to a few seconds, regardless of the file system size.
By default, VxFS file systems log file transactions before they are committed to
disk, reducing time spent checking and repairing file systems after the system is
halted unexpectedly.
Extents
An extent is a contiguous area of storage in a computer file system, reserved for
a file. When starting to write to a file, a whole extent is allocated. When writing
to the file again, the data continues where the previous write left off. This reduces
or eliminates file fragmentation.
Since VxFS is an extent-based file system, addressing is done through extents
(which can consist of multiple blocks) rather than in single blocks segments.
Extents can therefore enhance file system throughput.
File system disk layouts

The disk layout is the way file system information is stored on disk. On VxFS,
several disk layout versions, numbered 1 through 7, were created to support
various new features and specific UNIX environments. Currently, only the Version
6 and 7 disk layouts can be created and mounted.

VxFS includes the following features:
Introducing Veritas File System 13
■ Extent-based allocation
Extents allow disk I/O to take place in units of multiple blocks if storage is
allocated in consecutive blocks.
■ Extent attributes
Extent attributes are the extent allocation policies associated with a file.
■ Fast file system recovery
VxFS provides fast recovery of a file system from system failure.
■ Extended mount options
The VxFS file system supports extended mount options to specify enhanced
data integrity modes, enhanced performance modes, temporary file system
modes, improved synchronous writes, and large file sizes.
■ Enhanced data integrity modes
VxFS avoids the problem of uninitialized data appearing in a file by waiting
until the data has been flushed to disk before updating the new file size to disk.
■ Enhanced performance mode
VxFS provides mount options to improve performance.
■ Modes of temporary file systems
VxFS supplies an option to allow users to achieve higher performance on
temporary file sytems by delaying the logging for most operations.
■ Improved synchronous writes
VxFS provides superior performance for synchronous write applications.
■ Large files and file systems support
VxFS supports files larger than two terabytes and large file systems up to 256
terabytes.
■ Access control lists (ACLs)
An Access Control List (ACL) stores a series of entries that identify specific
users or groups and their access privileges for a directory or file.
■ Storage Checkpoints
Backup and restore applications can leverage Storage Checkpoint, a disk- and
I/O-efficient copying technology for creating periodic frozen images of a file
system.
■ Online backup
VxFS provides online data backup using the snapshot feature.
■ Quotas
VxFS supports quotas, which allocate per-user and per-group quotas and limit
the use of two principal resources: files and data blocks.
■ Cluster File System

Clustered file systems are an extension of VxFS that support concurrent direct
media access from multiple systems.
■ Improved database performance
Databases can be created on the character devices to achieve the same
performance as databases created on raw disks.
■ Cross-platform data sharing
Cross-platform data sharing allows data to be serially shared among
heterogeneous systems where each system has direct access to the physical
devices that hold the data.
■ File Change Log
The VxFS File Change Log tracks changes to files and directories in a file
system.
■ Multi-volume support
The multi-volume support feature allows several volumes to be represented
by a single logical object.
■ Dynamic Storage Tiering
The Dynamic Storage Tiering (DST) option allows you to configure policies
that automatically relocate files from one volume to another, or relocate files
by running file relocation commands, which can improve performance for
applications that access specific types of files.
Note: VxFS supports all UFS file system features and facilities except for linking,
removing, or renaming “.” and “..” directory entries. These operations may
disrupt file system operations.
Extent-based allocation
Disk space is allocated in 512-byte sectors to form logical blocks. VxFS supports
logical block sizes of 1024, 2048, 4096, and 8192 bytes. The default block size
is 1K. For file systems up to 4 TB, the block size is 1K. 2K for file systems up to
8TB, 4K for file systems up to 16 TB, and 8K for file systems beyond this size.
An extent is defined as one or more adjacent blocks of data within the file system.
An extent is presented as an address-length pair, which identifies the starting
block address and the length of the extent (in file system or logical blocks). VxFS
allocates storage in groups of extents rather than a block at a time.
Extents allow disk I/O to take place in units of multiple blocks if storage is allocated
in consecutive blocks. For sequential I/O, multiple block operations are
considerably faster than block-at-a-time operations; almost all disk drives accept
I/O operations of multiple blocks.
Extent allocation only slightly alters the interpretation of addressed blocks from
the inode structure compared to block based inodes. A VxFS inode references 10
direct extents, each of which are pairs of starting block addresses and lengths in
blocks.
The VxFS inode supports different types of extents, namely ext4 and typed. Inodes
with ext4 extents also point to two indirect address extents, which contain the
addresses of first and second extents:
first Used for single indirection. Each entry in the extent indicates the
starting block number of an indirect data extent
second Used for double indirection. Each entry in the extent indicates the
starting block number of a single indirect address extent.
Each indirect address extent is 8K long and contains 2048 entries. All indirect
data extents for a file must be the same size; this size is set when the first indirect
data extent is allocated and stored in the inode. Directory inodes always use an
8K indirect data extent size. By default, regular file inodes also use an 8K indirect
data extent size that can be altered with vxtunefs; these inodes allocate the
indirect data extents in clusters to simulate larger extents.
Typed extents
VxFS has an inode block map organization for indirect extents known as typed
extents. Each entry in the block map has a typed descriptor record containing a
type, offset, starting block, and number of blocks.
Indirect and data extents use this format to identify logical file offsets and physical
disk locations of any given extent.
The extent descriptor fields are defined as follows:
type Identifies uniquely an extent descriptor record and defines the record's
length and format.
offset Represents the logical file offset in blocks for a given descriptor. Used
to optimize lookups and eliminate hole descriptor entries.
starting block Is the starting file system block of the extent.
number of blocks Is the number of contiguous blocks in the extent.
Typed extents have the following characteristics:

■ Indirect address blocks are fully typed and may have variable lengths up to a
maximum and optimum size of 8K. On a fragmented file system, indirect
extents may be smaller than 8K depending on space availability. VxFS always
tries to obtain 8K indirect extents but resorts to smaller indirects if necessary.
■ Indirect data extents are variable in size to allow files to allocate large,
contiguous extents and take full advantage of optimized I/O in VxFS.
■ Holes in sparse files require no storage and are eliminated by typed records.
A hole is determined by adding the offset and length of a descriptor and
comparing the result with the offset of the next record.
■ While there are no limits on the levels of indirection, lower levels are expected
in this format since data extents have variable lengths.
■ This format uses a type indicator that determines its record format and content
and accommodates new requirements and functionality for future types.
The current typed format is used on regular files and directories only when
indirection is needed. Typed records are longer than the previous format and
require less direct entries in the inode. Newly created files start out using the old
format, which allows for ten direct extents in the inode. The inode's block map is
converted to the typed format when indirection is needed to offer the advantages
of both formats.
Extent attributes
VxFS allocates disk space to files in groups of one or more extents. VxFS also
allows applications to control some aspects of the extent allocation. Extent
attributes are the extent allocation policies associated with a file.
The setext and getext commands allow the administrator to set or view extent
attributes associated with a file, as well as to preallocate space for a file.
See the setext(1) and getext(1) manual pages.
The vxtunefs command allows the administrator to set or view the default indirect
data extent size.
See the vxtunefs(1M) manual page.
Fast file system recovery

Most file systems rely on full structural verification by the fsck utility as the only
means to recover from a system failure. For large disk configurations, this involves
a time-consuming process of checking the entire structure, verifying that the file
system is intact, and correcting any inconsistencies. VxFS provides fast recovery
with the VxFS intent log and VxFS intent log resizing features.
VxFS intent log

VxFS reduces system failure recovery times by tracking file system activity in the
VxFS intent log. This feature records pending changes to the file system structure
in a circular intent log. The intent log recovery feature is not readily apparent to
users or a system administrator except during a system failure. During system
failure recovery, the VxFS fsck utility performs an intent log replay, which scans
the intent log and nullifies or completes file system operations that were active
when the system failed. The file system can then be mounted without completing
a full structural check of the entire file system. Replaying the intent log may not
completely recover the damaged file system structure if there was a disk hardware
failure; hardware problems may require a complete system check using the fsck
utility provided with VxFS.
See “The log option and data integrity” on page 18.
VxFS intent log resizing

The VxFS intent log is allocated when the file system is first created. The size of
the intent log is based on the size of the file system—the larger the file system,
the larger the intent log. The maximum default intent log size for disk layout
Versions 4 and 5 is 16 megabytes. The maximum default intent log size for disk
layout Version 6 and 7 is 64 megabytes.
With the Version 6 and 7 disk layouts, you can dynamically increase or decrease
the intent log size using the logsize option of the fsadm command. Increasing the
size of the intent log can improve system performance because it reduces the
number of times the log wraps around. However, increasing the intent log size
can lead to greater times required for a log replay if there is a system failure.
Note: Inappropriate sizing of the intent log can have a negative impact on system
performance.
See the mkfs_vxfs(1M) and the fsadm_vxfs(1M) manual pages.
Extended mount options

The VxFS file system supports the following extended mount options:
■ Enhanced data integrity modes
■ Enhanced performance modes
■ Temporary file system modes
■ Improved synchronous writes
■ Large file sizes

See the mount_vxfs(1M) manual page.
Enhanced data integrity modes

For most UNIX file systems, including VxFS, the default mode for writing to a file
is delayed, or buffered, meaning that the data to be written is copied to the file
system cache and later flushed to disk.
A delayed write provides much better performance than synchronously writing
the data to disk. However, in the event of a system failure, data written shortly
before the failure may be lost since it was not flushed to disk. In addition, if space
was allocated to the file as part of the write request, and the corresponding data
was not flushed to disk before the system failure occurred, uninitialized data can
appear in the file.
For the most common type of write, delayed extending writes (a delayed write
that increases the file size), VxFS avoids the problem of uninitialized data
appearing in the file by waiting until the data has been flushed to disk before
updating the new file size to disk. If a system failure occurs before the data has
been flushed to disk, the file size has not yet been updated to be uninitialized data,
thus no uninitialized data appears in the file. The unused blocks that were allocated
are reclaimed.
The blkclear option and data integrity

In environments where performance is more important than absolute data
integrity, the preceding situation is not of great concern. However, VxFS supports
environments that emphasize data integrity by providing the mount -o blkclear
option that ensures uninitialized data does not appear in a file.
The closesync option and data integrity

VxFS provides the mount -o mincache=closesync option, which is useful in desktop
environments with users who are likely to shut off the power on machines without
halting them first. In closesync mode, only files that are written during the system
crash or shutdown can lose data. Any changes to a file are flushed to disk when
the file is closed.
The log option and data integrity

File systems are typically asynchronous in that structural changes to the file
system are not immediately written to disk, which provides better performance.
However, recent changes made to a system can be lost if a system failure occurs.
Specifically, attribute changes to files and recently created files may disappear.
The mount -o log intent logging option guarantees that all structural changes to
the file system are logged to disk before the system call returns to the application.
With this option, the rename(2) system call flushes the source file to disk to
guarantee the persistence of the file data before renaming it. The rename() call is
also guaranteed to be persistent when the system call returns. The changes to file
system data and metadata caused by the fsync(2) and fdatasync(2) system calls
are guaranteed to be persistent once the calls return.
Enhanced performance mode

VxFS has mount options that improve performance, such as delaylog.
The delaylog option and enhanced performance

The default VxFS logging mode, mount -o delaylog, increases performance by
delaying the logging of some structural changes. However, it does not provide the
equivalent data integrity as the previously described modes.however, because
recent changes may be lost during a system failure. This option provides at least
the same level of data accuracy that traditional UNIX file systems provide for
system failures, along with fast file system recovery.
Modes of temporary file systems

On most UNIX systems, temporary file system directories, such as /tmp and
/usr/tmp, often hold files that do not need to be retained when the system reboots.
The underlying file system does not need to maintain a high degree of structural
integrity for these temporary directories. VxFS provides a mount -o tmplog option,
which allows the user to achieve higher performance on temporary file systems
by delaying the logging of most operations.
Improved synchronous writes

VxFS provides superior performance for synchronous write applications. The
mount -o datainlog option greatly improves the performance of small synchronous
writes.
The mount -o convosync=dsync option improves the performance of applications
that require synchronous data writes but not synchronous inode time updates.
Warning: The use of the -o convosync=dsync option violates POSIX semantics.

Support for large files

With VxFS, you can create, mount, and manage file systems containing large files
(files larger than two terabytes).
Warning: Some applications and utilities may not work on large files.
Access Control Lists

An Access Control List (ACL) stores a series of entries that identify specific users
or groups and their access privileges for a directory or file. A file may have its
own ACL or may share an ACL with other files. ACLs have the advantage of
specifying detailed access permissions for multiple users and groups. ACLs are
supported on cluster file systems.
See the getfacl(1) and setfacl(1) manual pages.
Storage Checkpoints
To increase availability, recoverability, and performance, Veritas File System
offers on-disk and online backup and restore capabilities that facilitate frequent
and efficient backup strategies. Backup and restore applications can leverage a
Storage Checkpoint, a disk- and I/O-efficient copying technology for creating
periodic frozen images of a file system. Storage Checkpoints present a view of a
file system at a point in time, and subsequently identifies and maintains copies
of the original file system blocks. Instead of using a disk-based mirroring method,
Storage Checkpoints save disk space and significantly reduce I/O overhead by
using the free space pool available to a file system.
Storage Checkpoint functionality is separately licensed.
Online backup
VxFS provides online data backup using the snapshot feature. An image of a
mounted file system instantly becomes an exact read-only copy of the file system
at a specific point in time. The original file system is called the snapped file system,
the copy is called the snapshot.
When changes are made to the snapped file system, the old data is copied to the
snapshot. When the snapshot is read, data that has not changed is read from the
snapped file system, changed data is read from the snapshot.
Backups require one of the following methods:
■ Copying selected files from the snapshot file system (using find and cpio)
■ Backing up the entire file system (using fscat)

■ Initiating a full or incremental backup (using vxdump)
See “About snapshot file systems” on page 93.
Quotas
VxFS supports quotas, which allocate per-user and per-group quotas and limit
the use of two principal resources: files and data blocks. You can assign quotas
for each of these resources. Each quota consists of two limits for each resource:
hard limit and soft limit.
The hard limit represents an absolute limit on data blocks or files. A user can
never exceed the hard limit under any circumstances.
The soft limit is lower than the hard limit and can be exceeded for a limited amount
of time. This allows users to exceed limits temporarily as long as they fall under
those limits before the allotted time expires.
See “About quota limits” on page 101.
Support for databases

Databases are usually created on file systems to simplify backup, copying, and
moving tasks and are slower compared to databases on raw disks.
Using Quick I/O for Databases feature with VxFS lets systems retain the benefits
of having a database on a file system without sacrificing performance. Quick I/O
creates regular, preallocated files to use as character devices. Databases can be
created on the character devices to achieve the same performance as databases
created on raw disks.
Treating regular VxFS files as raw devices has the following advantages for
databases:
■ Commercial database servers such as Oracle Server can issue kernel supported
asynchronous I/O calls on these pseudo devices but not on regular files.
Server can issue kernel
supported asynchronous I/O calls on these pseudo devices but not on regular
files.
■ read() and write() system calls issued by the database server can avoid the
acquisition and release of read/write locks inside the kernel that take place
on regular files.
■ VxFS can avoid double buffering of data already buffered by the database
server. This ability frees up resources for other purposes and results in better
performance.
■ Since I/O to these devices bypasses the system buffer cache, VxFS saves on
the cost of copying data between user space and kernel space when data is
read from or written to a regular file. This process significantly reduces CPU
time per I/O transaction compared to that of buffered I/O.
Cluster file systems

Clustered file systems are an extension of VxFS that support concurrent direct
media access from multiple systems. Cluster file system (CFS) employs a
master/slave protocol. All cluster file systems can read file data directly from a
shared disk. In addition, all systems can write “in-place” file data. Operations that
require changes to file system metadata, such as allocation, creation, and deletion,
can only be performed by the single primary file system node. To maintain file
system consistency, secondary nodes must send messages to the primary, and
the primary will perform the operations.
Installing VxFS and enabling the cluster feature does not create a cluster file
system configuration. File system clustering requires other Veritas products to
enable communication services and provide storage resources. These products
are packaged with VxFS in the Storage Foundation Cluster File System to provide
a complete clustering environment.
See the Veritas Storage Foundation Cluster File System Administrator's Guide.
To be a cluster mount, a file system must be mounted using the mount –o cluster
option. File systems mounted without the –o cluster option are termed local
mounts.
CFS functionality is separately licensed.
Cross-platform data sharing

Cross-platform data sharing (CDS) allows data to be serially shared among
heterogeneous systems where each system has direct access to the physical devices
that hold the data. This feature can be used only in conjunction with Veritas
Volume Manager (VxVM).
See the Veritas Storage Foundation Cross-Platform Data Sharing Administrator's
Guide.
File Change Log

The VxFS File Change Log (FCL) tracks changes to files and directories in a file
system. The File Change Log can be used by applications such as backup products,
webcrawlers, search and indexing engines, and replication software that typically
Veritas File System performance enhancements
scan an entire file system searching for modifications since a previous scan. FCL
functionality is a separately licensed feature.
See “About the File Change Log file” on page 109.
Multi-volume support
The multi-volume support (MVS) feature allows several volumes to be represented
by a single logical object. All I/O to and from an underlying logical volume is
directed by way of volume sets. This feature can be used only in conjunction with
VxVM. MVS functionality is a separately licensed feature.
See “About multi-volume support” on page 118.
Dynamic Storage Tiering

The Dynamic Storage Tiering (DST) option is built on multi-volume support
technology. Using DST, you can map more than one volume to a single file system.
You can then configure policies that automatically relocate files from one volume
to another, or relocate files by running file relocation commands. Having multiple
volumes lets you determine where files are located, which can improve
performance for applications that access specific types of files. DST functionality
is a separately licensed feature and is available with the VRTSfppm package.
See “About Dynamic Storage Tiering” on page 137.

Traditional file systems employ block-based allocation schemes that provide
adequate random access and latency for small files, but which limit throughput
for larger files. As a result, they are less than optimal for commercial
environments.
VxFS addresses this file system performance issue through an alternative
allocation method and increased user control over allocation, I/O, and caching
policies.
See “Using Veritas File System” on page 25.
VxFS provides the following performance enhancements:
■ Data synchronous I/O
■ Direct I/O and discovered direct I/O
■ Enhanced I/O performance
■ Caching advisories
■ Enhanced directory features

■ Explicit file alignment, extent size, and preallocation controls
■ Tunable I/O parameters
■ Tunable indirect data extent size
■ Integration with VxVM™
■ Support for large directories
Note: VxFS reduces the file lookup time in directories with an extremely large
number of files.
About enhanced I/O performance

VxFS provides enhanced I/O performance by applying an aggressive I/O clustering
policy, integrating with VxVM, and allowing application specific parameters to
be set on a per-file system basis.
Enhanced I/O clustering

I/O clustering is a technique of grouping multiple I/O operations together for
improved performance. VxFS I/O policies provide more aggressive clustering
processes than other file systems and offer higher I/O throughput when using
large files. The resulting performance is comparable to that provided by raw disk.
VxVM integration
VxFS interfaces with VxVM to determine the I/O characteristics of the underlying
volume and perform I/O accordingly. VxFS also uses this information when using
mkfs to perform proper allocation unit alignments for efficient I/O operations
from the kernel. VxFS also uses this information when using mkfs to perform
proper allocation unit alignments for efficient I/O operations from the kernel.
As part of VxFS/VxVM integration, VxVM exports a set of I/O parameters to
achieve better I/O performance. This interface can enhance performance for
different volume configurations such as RAID-5, striped, and mirrored volumes.
Full stripe writes are important in a RAID-5 volume for strong I/O performance.
VxFS uses these parameters to issue appropriate I/O requests to VxVM.
Application-specific parameters
You can also set application specific parameters on a per-file system basis to
improve I/O performance.
Using Veritas File System
■ Discovered Direct I/O

All sizes above this value would be performed as direct I/O.
■ Maximum Direct I/O Size
This value defines the maximum size of a single direct I/O.
See the vxtunefs(1M) and tunefstab(4) manual pages.

There are three main methods to use, manage, modify, and tune
ProductNameLong;:
■ See “Veritas Enterprise Administrator Graphical User Interface” on page 25.
■ See “Online system administration” on page 26.
■ See “Application program interface” on page 27.
Veritas Enterprise Administrator Graphical User Interface

Enterprise Administrator (VEA) is a GUI-based application using the Java™
technology that consists of a server and a client. The server runs on a UNIX system
that is running Volume Manager and VxFS. The client runs on any platform that
supports the Java Runtime Environment.
You can perform the following administrative functions on local or remote systems
using the GUI:
■ Create a file system on a volume
■ Create a file system on a volume set
■ Remove a file system from the file system table
■ Mount and unmounting a file system
■ Defragment a file system
■ Monitor file system capacity
■ Create a snapshot copy of a file system
■ Check a file system
■ View file system properties
■ Unmount a file system from a cluster node
■ Remove resource information for a cluster file system
■ Maintain the File Change Log
■ Maintain Storage Checkpoints

■ Use multi-volume file systems
■ Set intent log options
Online system administration

VxFS provides command line interface (CLI) operations that are described
throughout this guide and in manual pages.
VxFS allows you to run a number of administration tasks while the file system is
online. Two of the more important tasks include:
■ Deframentation
■ File system resizing
About defragmentation
Free resources are initially aligned and allocated to files in an order that provides
optimal performance. On an active file system, the original order of free resources
is lost over time as files are created, removed, and resized. The file system is
spread farther along the disk, leaving unused gaps or fragments between areas
that are in use. This process is known as fragmentation and leads to degraded
performance because the file system has fewer options when assigning a free
extent to a file (a group of contiguous data blocks).
VxFS provides the online administration utility fsadm to resolve the problem of
fragmentation.
The fsadm utility defragments a mounted file system by performing the following
actions:
■ Removing unused space from directories
■ Making all small files contiguous
■ Consolidating free blocks for file system use
This utility can run on demand and should be scheduled regularly as a cron job.
About file system resizing

A file system is assigned a specific size as soon as it is created; the file system may
become too small or too large as changes in file system usage take place over time.
VxFS is capable of increasing or decreasing the file system size while in use. Many
competing file systems can not do this. The VxFS utility fsadm can expand or
shrink a file system without unmounting the file system or interrupting user
productivity. However, to expand a file system, the underlying device on which

it is mounted must be expandable.
VxVM facilitates expansion using virtual disks that can be increased in size while
in use. The VxFS and VxVM packages complement each other to provide online
expansion capability. Use the vxresize command when resizing both the volume
and the file system. The vxresize command guarantees that the file system
shrinks or grows along with the volume. Do not use the vxassist and fsadm_vxfs
commands for this purpose.
See the vxresize(1M) manual page.
See the Veritas Volume Manager Administrator's Guide.
Application program interface

Veritas File System Developer's Kit (SDK) provides developers with the information
necessary to use the application programming interfaces (APIs) to modify and
tune various features and components of File System.
See the Veritas File System Programmer's Reference Guide.
VxFS conforms to the System V Interface Definition (SVID) requirements and
supports user access through the Network File System (NFS). Applications that
require performance features not available with other file systems can take
advantage of VxFS enhancements.
Expanded application facilities

VxFS provides API functions frequently associated with commercial applications
that make it possible to perform the following actions:
■ Preallocate space for a file
■ Specify a fixed extent size for a file
■ Bypass the system buffer cache for file I/O
■ Specify the expected access pattern for a file
Because these functions are provided using VxFS-specific ioctl system calls, most
existing UNIX system applications do not use them. The VxFS-specific cp, cpio,
and mv utilities use the functions to preserve extent attributes and allocate space
more efficiently. The current attributes of a file can be listed using the getext or
VxFS-specific ls command. The functions can also improve performance for
custom applications. For portability reasons, these applications must check which
file system type they are using before using these functions.
Chapter 2
VxFS performance:
creating, mounting, and
tuning File Systems
■ mkfs command options
■ Choosing mount command options
■ Using kernel tunables
■ Monitoring free space
■ Tuning I/O
mkfs command options

When you create a file system, you can select a number of characteristics.
■ See “Block size” on page 29.
■ See “Intent log size” on page 30.
Block size
The unit of allocation in VxFS is a block. Unlike some other UNIX file systems,
VxFS does not make use of block fragments for allocation because storage is
allocated in extents that consist of one or more blocks.
30 VxFS performance: creating, mounting, and tuning File Systems
Choosing mount command options
You specify the block size when creating a file system by using the mkfs –o bsize
option. The block size cannot be altered after the file system is created. The
smallest available block size for VxFS is 1K, which is also the default block size.
Choose a block size based on the type of application being run. For example, if
there are many small files, a 1K block size may save space. For large file systems,
with relatively few files, a larger block size is more appropriate. Larger block sizes
use less disk space in file system overhead, but consume more space for files that
are not a multiple of the block size. The easiest way to judge which block sizes
provide the greatest system efficiency is to try representative system loads against
various sizes and pick the fastest. For most applications, it is best to use the default
values.
For 64-bit kernels, which support 32 terabyte file systems, the block size
determines the maximum size of the file system you can create. File systems up
to 4 TB require a 1K block size. For four to eight terabyte file systems, the block
size is 2K, For file systems between 8 and 16 TB, block size is 4K, and for greater
than 16 TB, the block size is 8K. If you specify the file system size when creating
a file system, the block size defaults to these values.
Intent log size

You specify the intent log size when creating a file system by using the mkfs –o
logsize option. With the Version 6 or 7 disk layout, you can dynamically increase
or decrease the intent log size using the log option of the fsadm command. The
mkfs utility uses a default intent log size 64 megabytes for disk layout Version 6.
The default size is sufficient for most workloads. If the system is used as an NFS
server or for intensive synchronous write workloads, performance may be
improved using a larger log size.
With larger intent log sizes, recovery time is proportionately longer and the file
system may consume more system resources (such as memory) during normal
operation.
There are several system performance benchmark suites for which VxFS performs
better with larger log sizes. As with block sizes, the best way to pick the log size
is to try representative system loads against various sizes and pick the fastest.

In addition to the standard mount mode (delaylog mode), VxFS provides the
following modes of operation:
■ log
■ delaylog
VxFS performance: creating, mounting, and tuning File Systems 31
■ tmplog
■ logsize
■ nodatainlog
■ blkclear
■ minicache
■ convosync
■ ioerror
■ largefiles|nolorgefiles
■ cio
Caching behavior can be altered with the mincache option, and the behavior of
O_SYNC and D_SYNC writes can be altered with the convosync option.
See the fcntl(2) manual page.
The delaylog and tmplog modes can significantly improve performance. The
improvement over log mode is typically about 15 to 20 percent with delaylog; with
tmplog, the improvement is even higher. Performance improvement varies,
depending on the operations being performed and the workload. Read/write
intensive loads should show less improvement, while file system structure
intensive loads (such as mkdir, create, and rename) may show over 100 percent
improvement. The best way to select a mode is to test representative system loads
against the logging modes and compare the performance results.
Most of the modes can be used in combination. For example, a desktop machine
might use both the blkclear and mincache=closesync modes.
The log mode

In log mode, all system calls other than write(2), writev(2), and pwrite(2) are
guaranteed to be persistent after the system call returns to the application.
The rename(2) system call flushes the source file to disk to guarantee the
persistence of the file data before renaming it. In both the log and delaylog modes,
the rename is also guaranteed to be persistent when the system call returns. This
benefits shell scripts and programs that try to update a file atomically by writing
the new file contents to a temporary file and then renaming it on top of the target
file.
The delaylog mode

The default logging mode is delaylog. In delaylog mode, the effects of most system
calls other than write(2), writev(2), and pwrite(2) are guaranteed to be persistent
approximately 15 to 20 seconds after the system call returns to the application.
Contrast this with the behavior of most other file systems in which most system
calls are not persistent until approximately 30 seconds or more after the call has
returned. Fast file system recovery works with this mode.
The rename(2) system call flushes the source file to disk to guarantee the
persistence of the file data before renaming it. In the log and delaylog modes, the
rename is also guaranteed to be persistent when the system call returns. This
benefits shell scripts and programs that try to update a file atomically by writing
the new file contents to a temporary file and then renaming it on top of the target
file.
The tmplog mode

In tmplog mode, the effects of system calls have persistence guarantees that are
similar to those in delaylog mode. In addition, enhanced flushing of delayed
extending writes is disabled, which results in better performance but increases
the chances of data being lost or uninitialized data appearing in a file that was
being actively written at the time of a system failure. This mode is only
recommended for temporary file systems. Fast file system recovery works with
this mode.
Note: The term “effects of system calls” refers to changes to file system data and
metadata caused by the system call, excluding changes to st_atime. See the stat(2)
manual page.
Persistence guarantees
In all logging modes, VxFS is fully POSIX compliant. The effects of the fsync(2)
and fdatasync(2) system calls are guaranteed to be persistent after the calls return.
The persistence guarantees for data or metadata modified by write(2), writev(2),
or pwrite(2) are not affected by the logging mount options. The effects of these
system calls are guaranteed to be persistent only if the O_SYNC, O_DSYNC,
VX_DSYNC, or VX_DIRECT flag, as modified by the convosync= mount option, has
been specified for the file descriptor.
The behavior of NFS servers on a VxFS file system is unaffected by the log and
tmplog mount options, but not delaylog. In all cases except for tmplog, VxFS
complies with the persistency requirements of the NFS v2 and NFS v3 standard.
Unless a UNIX application has been developed specifically for the VxFS file system
in log mode, it expects the persistence guarantees offered by most other file
systems and experiences improved robustness when used with a VxFS file system
mounted in delaylog mode. Applications that expect better persistence guarantees
than that offered by most other file systems can benefit from the log, mincache=,
and closesync mount options. However, most commercially available applications
work well with the default VxFS mount options, including the delaylog mode.
The logiosize mode

The logiosize=size option enhances the performance of storage devices that
employ a read-modify-write feature. If you specify logiosize when you mount a
file system, VxFS writes the intent log in the least size bytes or a multiple of size
bytes to obtain the maximum performance from such devices.
See mount_vxfs(1m) manual page.
The values for size can be 512, 1024, 2048, 4096, or 8192.
The nodatainlog mode

Use the nodatainlog mode on systems with disks that do not support bad block
revectoring. Usually, a VxFS file system uses the intent log for synchronous writes.
The inode update and the data are both logged in the transaction, so a synchronous
write only requires one disk write instead of two. When the synchronous write
returns to the application, the file system has told the application that the data
is already written. If a disk error causes the metadata update to fail, then the file
must be marked bad and the entire file is lost.
If a disk supports bad block revectoring, then a failure on the data update is
unlikely, so logging synchronous writes should be allowed. If the disk does not
support bad block revectoring, then a failure is more likely, so the nodatainlog
mode should be used.
A nodatainlog mode file system is approximately 50 percent slower than a standard
mode VxFS file system for synchronous writes. Other operations are not affected.
The blkclear mode

The blkclear mode is used in increased data security environments. The blkclear
mode guarantees that uninitialized storage never appears in files. The increased
integrity is provided by clearing extents on disk when they are allocated within
a file. This mode does not affect extending writes. A blkclear mode file system is
approximately 10 percent slower than a standard mode VxFS file system,
depending on the workload.
The mincache mode

The mincache mode has the following suboptions:
■ mincache=closesync
■ mincache=direct
■ mincache=dsync
■ mincache=unbuffered
■ mincache=tmpcache
The mincache=closesync mode is useful in desktop environments where users
are likely to shut off the power on the machine without halting it first. In this
mode, any changes to the file are flushed to disk when the file is closed.
To improve performance, most file systems do not synchronously update data
and inode changes to disk. If the system crashes, files that have been updated
within the past minute are in danger of losing data. With the mincache=closesync
mode, if the system crashes or is switched off, only open files can lose data. A
mincache=closesync mode file system could be approximately 15 percent slower
than a standard mode VxFS file system, depending on the workload.
The following describes where to use the mincache modes:
■ The mincache=direct, mincache=unbuffered, and mincache=dsync modes are
used in environments where applications have reliability problems caused by
the kernel buffering of I/O and delayed flushing of non-synchronous I/O.
■ The mincache=direct and mincache=unbuffered modes guarantee that all
non-synchronous I/O requests to files are handled as if the VX_DIRECT or
VX_UNBUFFERED caching advisories had been specified.
■ The mincache=dsync mode guarantees that all non-synchronous I/O requests
to files are handled as if the VX_DSYNC caching advisory had been specified.
Refer to the vxfsio(7) manual page for explanations of VX_DIRECT,
VX_UNBUFFERED, and VX_DSYNC, as well as for the requirements for direct
I/O.
■ The mincache=direct, mincache=unbuffered, and mincache=dsync modes also
flush file data on close as mincache=closesync does.
Because the mincache=direct, mincache=unbuffered, and mincache=dsync modes
change non-synchronous I/O to synchronous I/O, throughput can substantially
degrade for small to medium size files with most applications. Since the
VX_DIRECT and VX_UNBUFFERED advisories do not allow any caching of data,
applications that normally benefit from caching for reads usually experience less
degradation with the mincache=dsync mode. mincache=direct and
mincache=unbuffered require significantly less CPU time than buffered I/O.
If performance is more important than data integrity, you can use the
mincache=tmpcache mode. The mincache=tmpcache mode disables special delayed
extending write handling, trading off less integrity for better performance. Unlike
the other mincache modes, tmpcache does not flush the file to disk the file is
closed. When the mincache=tmpcache option is used, bad data can appear in a
file that was being extended when a crash occurred.
The convosync mode

The convosync (convert osync) mode has the following suboptions:
■ convosync=closesync
Note: The convosync=closesync mode converts synchronous and data

synchronous writes to non-synchronous writes and flushes the changes to the
file to disk when the file is closed.
■ convosync=delay
■ convosync=direct
■ convosync=dsync
Note: The convosync=dsync option violates POSIX guarantees for synchronous

I/O.
■ convosync=unbuffered
The convosync=delay mode causes synchronous and data synchronous writes to
be delayed rather than to take effect immediately. No special action is performed
when closing a file. This option effectively cancels any data integrity guarantees
normally provided by opening a file with O_SYNC.
See the open(2), fcntl(2), and vxfsio(7) manual pages.
Warning: Be very careful when using the convosync=closesync or convosync=delay

mode because they actually change synchronous I/O into non-synchronous I/O.
Applications that use synchronous I/O for data reliability may fail if the system
crashes and synchronously written data is lost.
The convosync=dsync mode converts synchronous writes to data synchronous

writes.
As with closesync, the direct, unbuffered, and dsync modes flush changes to the
file to disk when it is closed. These modes can be used to speed up applications
that use synchronous I/O. Many applications that are concerned with data integrity
specify the O_SYNC fcntl in order to write the file data synchronously. However,
this has the undesirable side effect of updating inode times and therefore slowing
down performance. The convosync=dsync, convosync=unbuffered, and
convosync=direct modes alleviate this problem by allowing applications to take
advantage of synchronous writes without modifying inode times as well.
Before using convosync=dsync, convosync=unbuffered, or convosync=direct,
make sure that all applications that use the file system do not require synchronous
inode time updates for O_SYNC writes.
The ioerror mode

This mode sets the policy for handling I/O errors on a mounted file system. I/O
errors can occur while reading or writing file data or metadata. The file system
can respond to these I/O errors either by halting or by gradually degrading. The
ioerror option provides five policies that determine how the file system responds
to the various errors. All policies limit data corruption, either by stopping the file
system or by marking a corrupted inode as bad.
The policies are the following:
■ disable
■ nodisable
■ wdisable
■ mwdisable
■ mdisable
The disable policy

If disable is selected, VxFS disables the file system after detecting any I/O error.
You must then unmount the file system and correct the condition causing the I/O
error. After the problem is repaired, run fsck and mount the file system again.
In most cases, replay fsck to repair the file system. A full fsck is required only
in cases of structural damage to the file system's metadata. Select disable in
environments where the underlying storage is redundant, such as RAID-5 or
mirrored disks.
The nodisable policy

If nodisable is selected, when VxFS detects an I/O error, it sets the appropriate
error flags to contain the error, but continues running. Note that the degraded
condition indicates possible data or metadata corruption, not the overall
performance of the file system.
For file data read and write errors, VxFS sets the VX_DATAIOERR flag in the
super-block. For metadata read errors, VxFS sets the VX_FULLFSCK flag in the
super-block. For metadata write errors, VxFS sets the VX_FULLFSCK and
VX_METAIOERR flags in the super-block and may mark associated metadata as
bad on disk. VxFS then prints the appropriate error messages to the console.
You should stop the file system as soon as possible and repair the condition causing
the I/O error. After the problem is repaired, run fsck and mount the file system
again. Select nodisable if you want to implement the policy that most closely
resembles the error handling policy of the previous VxFS release.
The wdisable and mwdisable policies

If wdisable (write disable) or mwdisable (metadata-write disable) is selected, the
file system is disabled or degraded, depending on the type of error encountered.
Select wdisable or mwdisable for environments where read errors are more likely
to persist than write errors, such as when using non-redundant storage. mwdisable
is the default ioerror mount option for local mounts.
The mdisable policy

If mdisable (metadata disable) is selected, the file system is disabled if a metadata
read or write fails. However the file system continues to operate if the failure is
confined to data extents. mdisable is the default ioerror mount option for cluster
mounts.
The largefiles|nolargefiles option

The section includes the following topics :
■ See “Creating a file system with large files” on page 38.
■ See “Mounting a file system with large files” on page 38.
■ See “Managing a file system with large files” on page 38.
VxFS supports files larger than two terabytes. Files larger than 32 terabytes can
be created only on 64-bit kernel operating systems and on a Veritas Volume
Manager volume.
Note: Applications and utilities such as backup may experience problems if they
are not aware of large files. In such a case, create your file system without large
file capability.
Creating a file system with large files

To create a file system with a file capability, type the following command:
# mkfs -F vxfs -o largefiles special_device size
Specifying largefiles sets the largefiles flag. This lets the file system to hold files
that are two terabytes or larger. This is the default option.
To clear the flag and prevent large files from being created, type the following
command:
# mkfs -F vxfs -o nolargefiles special_device size
The largefiles flag is persistent and stored on disk.
Mounting a file system with large files

If a mount succeeds and nolargefiles is specified, the file system cannot contain
or create any large files. If a mount succeeds and largefiles is specified, the file
system may contain and create large files.
The mount command fails if the specified largefiles|nolargefiles option does not
match the on-disk flag.
The mount command defaults to match the current setting of the on-disk flag if
specified without the largefiles or nolargefiles option, so it's best not to specify
either option. After a file system is mounted, you can use the fsadm utility to
change the large files option.
Managing a file system with large files

Managing a file system with large files includes the following tasks:
■ Determining the current status of the large files flag
■ Switching capabilities on a mounted file system
■ Switching capabilities on an unmounted file system
To determine the current status of the largefiles flag, type either of the following
commands:
# mkfs -F vxfs -m special_device

# fsadm -F vxfs mount_point | special_device
To switch capabilities on a mounted file system, type the following command;
# fsadm -F vxfs -o [no]largefiles mount_point
To switch capabilities on an unmounted file system, type the following command:
# fsadm -F vxfs -o [no]largefiles special_device
You cannot change a file system to nolargefiles if it holds large files.

See the mount_vxfs(1M), fsadm_vxfs(1M), and mkfs_vxfs(1M) manual pages.
The cio option

The cio (Concurent I/O) option specifies the file system to be mounted for
concurrent readers and writers. Concurrent I/O is a licensed feature of VxFS. If
cio is specified, but the feature is not licensed, the mount command prints an error
message and terminates the operation without mounting the file system. The cio
option cannot be disabled through a remount. To disable the cio option, the file
system must be unmounted and mounted again without the cio option.
Combining mount command options

Although mount options can be combined arbitrarily, some combinations do not
make sense. The following examples provide some common and reasonable mount
option combinations.
To mount a desktop file system using options, type the following:
# mount -F vxfs -o log,mincache=closesync /dev/dsk/c1t3d0s1 /mnt
This guarantees that when a file is closed, its data is synchronized to disk and
cannot be lost. Thus, after an application has exited and its files are closed, no
data is lost even if the system is immediately turned off.
To mount a temporary file system or to restore from backup, type the following:
# mount -F vxfs -o tmplog,convosync=delay,mincache=tmpcache \

/dev/dsk/c1t3d0s1 /mnt
This combination might be used for a temporary file system where performance
is more important than absolute data integrity. Any O_SYNC writes are performed
as delayed writes and delayed extending writes are not handled. This could result
in a file that contains corrupted data if the system crashes. Any file written 30
seconds or so before a crash may contain corrupted data or be missing if this
mount combination is in effect. However, such a file system does significantly
Using kernel tunables
less disk writes than a log file system, and should have significantly better
performance, depending on the application.
To mount a file system for synchronous writes, type the following:
# mount -F vxfs -o log,convosync=dsync /dev/dsk/c1t3d0s1 /mnt
This combination can be used to improve the performance of applications that

perform O_SYNC writes, but only require data synchronous write semantics.
Performance can be significantly improved if the file system is mounted using
convosync=dsync without any loss of data integrity.

This section describes the following kernel tunable parameters in VxFS:
■ Tuning inode table size
■ vx_maxlink
■ Veritas Volume Manager maximum I/O size
Tuning inode table size

Tuning the internal inode table size includes the following tasks:
■ Increasing the internal inode table size
■ Changing the size of the directory name lookup cache
VxFS caches inodes in an inode table. The tunable for VxFS to determine the
number of entries in its inode table is vxfs_ninode.
VxFS uses the value of vxfs_ninode in /etc/system as the number of entries in the
VxFS inode table. By default, the file system uses a value of vxfs_ninode, which
is computed based on system memory size.
To increase the internal inode table size
1 Open the /etc/system file.
2 Update the following line in the /etc/system file:
set vxfs:vxfs_ninode = new_value
It may be necessary to tune the dnlc (directory name lookup cache) size to keep
the value within an acceptable range relative to vxfs_ninode. It must be within
80% of vxfs_ninode to avoid spurious ENFILE errors or excessive CPU
consumption, but must be more than 50% of vxfs_ninode to maintain good
performance. The variable ncsize determines the size of dnlc. The default value
of ncsize is based on the kernel variable maxusers. It is computed at system boot

time. This value can be changed by making an entry in the /etc/system file:
set ncsize = new_value
The new ncsize is effective after you reboot the system.
vx_maxlink
The VxFS vx_maxlink tunable determines the number of sub-directories that can
be created under a directory.
A VxFS file system obtains the value of vx_maxlink from the system configuration
file /etc/system. By default, vx_maxlink is 32K. To change the computed value of
vx_maxlink, you can add an entry to the system configuration file. For example:
set vxfs:vx_maxlink = 65534
sets vx_maxlink to the maximum number of sub-directories. Valid values are 1

to 65534 (FFFE hexadecimal). Changes to vx_maxlink take effect after rebooting.
Veritas Volume Manager maximum I/O size

When using VxFS with Veritas Volume Manager (VxVM), VxVM by default breaks
up I/O requests larger than 256K. When using striping, to optimize performance,
the file system issues I/O requests that are up to a full stripe in size. If the stripe
size is larger than 256K, those requests are broken up.
vol_maxio
The vol_maxio parameter controls the maximum size of logical I/O operations
that can be performed without breaking up a request. Logical I/O requests larger
than this value are broken up and performed synchronously. Physical I/Os are
broken up based on the capabilities of the disk device and are unaffected by
changes to the vol_maxio logical request limit.
Raising the vol_maxio limit can cause problems if the size of an I/O requires more
memory or kernel mapping space than exists. The recommended maximum for
vol_maxio is 20% of the smaller of physical memory or kernel virtual memory. It
is not advisable to go over this limit. Within this limit, you can generally obtain
the best results by setting vol_maxio to the size of your largest stripe. This applies
to both RAID-0 striping and RAID-5 striping.
To increase the value of vol_maxio, add an entry to /etc/system (after the entry
forceload:drv/vxio) and reboot for the change to take effect. For example, the
following line sets the maximum I/O size to 16 MB:
Monitoring free space
set vxio:vol_maxio=32768
This parameter is in 512-byte sectors and is stored as a 16-bit number, so it cannot

be larger than 65535.
Monitoring free space

In general, VxFS works best if the percentage of free space in the file system does
not get below 10 percent. This is because file systems with 10 percent or more
free space have less fragmentation and better extent allocation. Regular use of
the df command to monitor free space is desirable.
See the df_vxfs(1M) manual page.
Full file systems may have an adverse effect on file system performance. Full file
systems should therefore have some files removed, or should be expanded.
See the fsadm_vxfs(1M) manual page.
Monitoring fragmentation
Fragmentation reduces performance and availability. Regular use of fsadm's
fragmentation reporting and reorganization facilities is therefore advisable.
The easiest way to ensure that fragmentation does not become a problem is to
schedule regular defragmentation runs using the cron command.
Defragmentation scheduling should range from weekly (for frequently used file
systems) to monthly (for infrequently used file systems). Extent fragmentation
should be monitored with fsadm command.
To determine the degree of fragmentation, use the following factors:
■ Percentage of free space in extents of less than 8 blocks in length
■ Percentage of free space in extents of less than 64 blocks in length
■ Percentage of free space in extents of length 64 blocks or greater
An unfragmented file system has the following characteristics:
■ Less than 1 percent of free space in extents of less than 8 blocks in length
■ Less than 5 percent of free space in extents of less than 64 blocks in length
■ More than 5 percent of the total file system size available as free extents in
lengths of 64 or more blocks
A badly fragmented file system has one or more of the following characteristics:
■ Greater than 5 percent of free space in extents of less than 8 blocks in length
Tuning I/O
■ More than 50 percent of free space in extents of less than 64 blocks in length
■ Less than 5 percent of the total file system size available as free extents in
lengths of 64 or more blocks
The optimal period for scheduling of extent reorganization runs can be determined
by choosing a reasonable interval, scheduling fsadm runs at the initial interval,
and running the extent fragmentation report feature of fsadm before and after
the reorganization.
The “before” result is the degree of fragmentation prior to the reorganization. If
the degree of fragmentation is approaching the figures for bad fragmentation,
reduce the interval between fsadm runs. If the degree of fragmentation is low,
increase the interval between fsadm runs.
The “after” result is an indication of how well the reorganizer has performed. The
degree of fragmentation should be close to the characteristics of an unfragmented
file system. If not, it may be a good idea to resize the file system; full file systems
tend to fragment and are difficult to defragment. It is also possible that the
reorganization is not being performed at a time during which the file system in
question is relatively idle.
Directory reorganization is not nearly as critical as extent reorganization, but
regular directory reorganization improves performance. It is advisable to schedule
directory reorganization for file systems when the extent reorganization is
scheduled. The following is a sample script that is run periodically at 3:00 A.M.
from cron for a number of file systems:
outfile=/usr/spool/fsadm/out.‘/bin/date +'%m%d'‘
for i in /home /home2 /project /db
do
/bin/echo "Reorganizing $i"
/bin/timex fsadm -F vxfs -e -E -s $i
/bin/timex fsadm -F vxfs -s -d -D $i
done > $outfile 2>&1
Tuning I/O
The performance of a file system can be enhanced by a suitable choice of I/O sizes
and proper alignment of the I/O requests based on the requirements of the
underlying special device. VxFS provides tools to tune the file systems.
Tuning I/O
Note: The following tunables and the techniques work on a per file system basis.
Use them judiciously based on the underlying device properties and characteristics
of the applications that use the file system.
Tuning VxFS I/O parameters

VxFS provides a set of tunable I/O parameters that control some of its behavior.
These I/O parameters are useful to help the file system adjust to striped or RAID-5
volumes that could yield performance superior to a single disk. Typically, data
streaming applications that access large files see the largest benefit from tuning
the file system.
VxVM queries
VxVM receives the following queries during configuration:
■ The file system queries VxVM to determine the geometry of the underlying
volume and automatically sets the I/O parameters.
Note: When using file systems in multiple volume sets, VxFS sets the VxFS
tunables based on the geometry of the first component volume (volume 0) in
the volume set.
■ The mkfs command queries VxVM when the file system is created to
automatically align the file system to the volume geometry. If the default
alignment from mkfs is not acceptable, the -o align=n option can be used to
override alignment information obtained from VxVM.
■ The mount command queries VxVM when the file system is mounted and
downloads the I/O parameters.
If the default parameters are not acceptable or the file system is being used without
VxVM, then the /etc/vx/tunefstab file can be used to set values for I/O parameters.
The mount command reads the /etc/vx/tunefstab file and downloads any
parameters specified for a file system. The tunefstab file overrides any values
obtained from VxVM. While the file system is mounted, any I/O parameters can
be changed using the vxtunefs command which can have tunables specified on
the command line or can read them from the /etc/vx/tunefstab file. For more
details,
See the vxtunefs(1M) and tunefstab(4) manual pages.
The vxtunefs command can be used to print the current values of the I/O
parameters.
Tuning I/O
To print the values, type the following command:
# vxtunefs -p mount_point
The following is an example tunefstab file:
/dev/vx/dsk/userdg/netbackup
read_pref_io=128k,write_pref_io=128k,read_nstream=4,write_nstream=4
/dev/vx/dsk/userdg/metasave
/dev/vx/dsk/userdg/solbuild
/dev/vx/dsk/userdg/solrelease
/dev/vx/dsk/userdg/solpatch
Tunable I/O parameters

Table 2-1 provides a list and description of these parameters.
Table 2-1 Tunable VxFS I/O parameters
Parameter Description
read_pref_io The preferred read request size. The file system uses this
in conjunction with the read_nstream value to determine
how much data to read ahead. The default value is 64K.
write_pref_io The preferred write request size. The file system uses this
in conjunction with the write_nstream value to determine
how to do flush behind on writes. The default value is 64K.
read_nstream The number of parallel read requests of size read_pref_io

to have outstanding at one time. The file system uses the
product of read_nstream multiplied by read_pref_io to
determine its read ahead size. The default value for
read_nstream is 1.
Tuning I/O
Table 2-1 Tunable VxFS I/O parameters (continued)
write_nstream The number of parallel write requests of size write_pref_io

to have outstanding at one time. The file system uses the
product of write_nstream multiplied by write_pref_io to
determine when to do flush behind on writes. The default
value for write_nstream is 1.
default_indir_ size On VxFS, files can have up to ten direct extents of variable
size stored in the inode. After these extents are used up,
the file must use indirect extents which are a fixed size
that is set when the file first uses indirect extents. These
indirect extents are 8K by default. The file system does
not use larger indirect extents because it must fail a write
and return ENOSPC if there are no extents available that
are the indirect extent size. For file systems containing
many large files, the 8K indirect extent size is too small.
The files that get into indirect extents use many smaller
extents instead of a few larger ones. By using this
parameter, the default indirect extent size can be
increased so large that files in indirects use fewer larger
extents. The tunable default_indir_size should be used
carefully. If it is set too large, then writes fail when they
are unable to allocate extents of the indirect extent size
to a file. In general, the fewer and the larger the files on
a file system, the larger the default_indir_size can be set.
This parameter should generally be set to some multiple
of the read_pref_io parameter. default_indir_size is not
applicable on Version 4 disk layouts.
discovered_direct_iosz Any file I/O requests larger than the

discovered_direct_iosz are handled as discovered direct
I/O. A discovered direct I/O is unbuffered similar to direct
I/O, but it does not require a synchronous commit of the
inode when the file is extended or blocks are allocated.
For larger I/O requests, the CPU time for copying the data
into the page cache and the cost of using memory to buffer
the I/O data becomes more expensive than the cost of
doing the disk I/O. For these I/O requests, using discovered
direct I/O is more efficient than regular I/O. The default
value of this parameter is 256K.
Tuning I/O
fcl_keeptime Specifies the minimum amount of time, in seconds, that

the VxFS File Change Log (FCL) keeps records in the log.
When the oldest 8K block of FCL records have been kept
longer than the value of fcl_keeptime, they are purged
from the FCL and the extents nearest to the beginning of
the FCL file are freed. This process is referred to as
“punching a hole.” Holes are punched in the FCL file in
8K chunks.
If the fcl_maxalloc parameter is set, records are purged

from the FCL if the amount of space allocated to the FCL
exceeds fcl_maxalloc, even if the elapsed time the records
have been in the log is less than the value of fcl_keeptime.
If the file system runs out of space before fcl_keeptime is
reached, the FCL is deactivated.
Either or both of the fcl_keeptime or fcl_maxalloc

parameters must be set before the File Change Log can be
activated. fcl_keeptime does not apply to disk layout
Versions 1 through 5.
fcl_maxalloc Specifies the maximum amount of space that can be

allocated to the VxFS File Change Log (FCL). The FCL file
is a sparse file that grows as changes occur in the file
system. When the space allocated to the FCL file reaches
the fcl_maxalloc value, the oldest FCL records are purged
from the FCL and the extents nearest to the beginning of
the FCL file are freed. This process is referred to as
“punching a hole.” Holes are punched in the FCL file in
8K chunks. If the file system runs out of space before
fcl_maxalloc is reached, the FCL is deactivated.
Either or both of the fcl_maxalloc or fcl_keeptime

parameters must be set before the File Change Log can be
activated. fcl_maxalloc does not apply to disk lay out
Versions 1 through 5.
Tuning I/O
fcl_winterval Specifies the time, in seconds, that must elapse before the
VxFS File Change Log (FCL) records a data overwrite, data
extending write, or data truncate for a file. The ability to
limit the number of repetitive FCL records for continuous
writes to the same file is important for file system
performance and for applications processing the FCL.
fcl_winterval is best set to an interval less than the
shortest interval between reads of the FCL by any
application. This way all applications using the FCL can
be assured of finding at least one FCL record for any file
experiencing continuous data changes.
fcl_winterval is enforced for all files in the file system.

Each file maintains its own time stamps, and the elapsed
time between FCL records is per file. This elapsed time
can be overridden using the VxFS FCL sync public API.
See the vxfs_fcl_sync(3) manual page.
fcl_winterval does not apply to disk layout Versions 1

through 5.
hsm_write_ prealloc For a file managed by a hierarchical storage management

(HSM) application, hsm_write_prealloc preallocates disk
blocks before data is migrated back into the file system.
An HSM application usually migrates the data back
through a series of writes to the file, each of which
allocates a few blocks. By setting hsm_write_prealloc
(hsm_write_prealloc=1), a sufficient number of disk blocks
are allocated on the first write to the empty file so that
no disk block allocation is required for subsequent writes.
This improves the write performance during migration.
The hsm_write_prealloc parameter is implemented outside

of the DMAPI specification, and its usage has limitations
depending on how the space within an HSM-controlled
file is managed. It is advisable to use hsm_write_prealloc
only when recommended by the HSM application
controlling the file system.
Tuning I/O
initial_extent_size Changes the default initial extent size. VxFS determines,

based on the first write to a new file, the size of the first
extent to be allocated to the file. Normally the first extent
is the smallest power of 2 that is larger than the size of
the first write. If that power of 2 is less than 8K, the first
extent allocated is 8K. After the initial extent, the file
system increases the size of subsequent extents with each
allocation.
See max_seqio_extent_size).
Since most applications write to files using a buffer size

of 8K or less, the increasing extents start doubling from
a small initial extent. initial_extent_size can change the
default initial extent size to be larger, so the doubling
policy starts from a much larger initial size and the file
system does not allocate a set of small extents at the start
of file. Use this parameter only on file systems that have
a very large average file size. On these file systems it
results in fewer extents per file and less fragmentation.
initial_extent_size is measured in file system blocks.
inode_aging_count Specifies the maximum number of inodes to place on an

inode aging list. Inode aging is used in conjunction with
file system Storage Checkpoints to allow quick restoration
of large, recently deleted files. The aging list is maintained
in first-in-first-out (fifo) order up to maximum number
of inodes specified by inode_aging_count. As newer inodes
are placed on the list, older inodes are removed to
complete their aging process. For best performance, it is
advisable to age only a limited number of larger files
before completion of the removal process. The default
maximum number of inodes to age is 2048.
inode_aging_size Specifies the minimum size to qualify a deleted inode for

inode aging. Inode aging is used in conjunction with file
system Storage Checkpoints to allow quick restoration of
large, recently deleted files. For best performance, it is
advisable to age only a limited number of larger files
before completion of the removal process. Setting the size
too low can push larger file inodes out of the aging queue
to make room for newly removed smaller file inodes.
Tuning I/O
max_direct_iosz The maximum size of a direct I/O request that are issued
by the file system. If a larger I/O request comes in, then
it is broken up into max_direct_iosz chunks. This
parameter defines how much memory an I/O request can
lock at once, so it should not be set to more than 20
percent of memory.
max_diskq Limits the maximum disk queue generated by a single file.

When the file system is flushing data for a file and the
number of pages being flushed exceeds max_diskq,
processes are blocked until the amount of data being
flushed decreases. Although this does not limit the actual
disk queue, it prevents flushing processes from making
the system unresponsive. The default value is 1 MB.
max_seqio_extent_size Increases or decreases the maximum size of an extent.

When the file system is following its default allocation
policy for sequential writes to a file, it allocates an initial
extent which is large enough for the first write to the file.
When additional extents are allocated, they are
progressively larger (the algorithm tries to double the size
of the file with each new extent) so each extent can hold
several writes worth of data. This is done to reduce the
total number of extents in anticipation of continued
sequential writes. When the file stops being written, any
unused space is freed for other files to use. Normally this
allocation stops increasing the size of extents at 2048
blocks which prevents one file from holding too much
unused space. max_seqio_extent_size is measured in file
system blocks.
qio_cache_enable Enables or disables caching on Quick I/O files. The default

behavior is to disable caching. To enable caching, set
qio_cache_enable to 1. On systems with large memories,
the database cannot always use all of the memory as a
cache. By enabling file system caching as a second level
cache, performance may be improved. If the database is
performing sequential scans of tables, the scans may run
faster by enabling file system caching so the file system
performs aggressive read-ahead on the files.
Tuning I/O
read_ahead The default for all VxFS read operations is to perform

sequential read ahead. You can specify the read_ahead
cache advisory to implement the VxFS enhanced read
ahead functionality. This allows read aheads to detect
more elaborate patterns (such as increasing or decreasing
read offsets or multithreaded file accesses) in addition to
simple sequential reads. You can specify the following
values for read_ahead:
0—Disables read ahead functionality
1—Retains traditional sequential read ahead behavior
2—Enables enhanced read ahead for all reads
The default is 1—VxFS detects only sequential patterns.
read_ahead detects patterns on a per-thread basis, up to

a maximum determined by vx_era_nthreads parameter.
The default number of threads is 5, but you can change
the default value by setting the vx_era_nthreads
parameter in the /etc/system configuration file.
Tuning I/O
write_throttle The write_throttle parameter is useful in special situations

where a computer system has a combination of a large
amount of memory and slow storage devices. In this
configuration, sync operations (such as fsync()) may
take long enough to complete that a system appears to
hang. This behavior occurs because the file system is
creating dirty pages (in-memory updates) faster than they
can be asynchronously flushed to disk without slowing
system performance.
Lowering the value of write_throttle limits the number

of dirty pages per file that a file system generates before
flushing the pages to disk. After the number of dirty pages
for a file reaches the write_throttle threshold, the file
system starts flushing pages to disk even if free memory
is still available.
The default value of write_throttle is zero, which puts no

limit on the number of dirty pages per file. If non-zero,
VxFS limits the number of dirty pages per file to
write_throttle pages.
The default value typically generates a large number of

dirty pages, but maintains fast user writes. Depending on
the speed of the storage device, if you lower write_throttle,
user write performance may suffer, but the number of
dirty pages is limited, so sync operations completes much
faster.
Because lowering write_throttle may in some cases delay

write requests (for example, lowering write_throttle may
increase the file disk queue to the max_diskq value,
delaying user writes until the disk queue decreases), it is
advisable not to change the value of write_throttle unless
your system has a combination of large physical memory
and slow storage devices.
File system tuning guidelines

If the file system is being used with VxVM, it is advisable to let the VxFS I/O
parameters be set to default values based on the volume geometry.
Tuning I/O
Note: VxFS does not query VxVM with multiple volume sets. To improve I/O
performance when using multiple volume sets, use the vxtunefs command.
If the file system is being used with a hardware disk array or volume manager
other than VxVM, try to align the parameters to match the geometry of the logical
disk. With striping or RAID-5, it is common to set read_pref_io to the stripe unit
size and read_nstream to the number of columns in the stripe. For striped arrays,
use the same values for write_pref_io and write_nstream, but for RAID-5 arrays,
set write_pref_io to the full stripe size and write_nstream to 1.
For an application to do efficient disk I/O, it should use the following formula to
issue read requests:
■ read requests = read_nstream x by read_pref_io
Generally, any multiple or factor of read_nstream multiplied by read_pref_io
should be a good size for performance. For writing, the same rule of thumb applies
to the write_pref_io and write_nstream parameters. When tuning a file system,
the best thing to do is try out the tuning parameters under a real life workload.
If an application is doing sequential I/O to large files, it should try to issue requests
larger than the discovered_direct_iosz. This causes the I/O requests to be
performed as discovered direct I/O requests, which are unbuffered like direct I/O
but do not require synchronous inode updates when extending the file. If the file
is larger than can fit in the cache, using unbuffered I/O avoids removing useful
data out of the cache and lessens CPU overhead.
Tuning I/O
Chapter 3
Extent attributes
■ About extent attributes
■ Commands related to extent attributes
About extent attributes

The Veritas File System (VxFS) allocates disk space to files in groups of one or
more adjacent blocks called extents. VxFS defines an application interface that
allows programs to control various aspects of the extent allocation for a given
file. The extent allocation policies associated with a file are referred to as extent
attributes.
The VxFS getext and setext commands let you view or manipulate file extent
attributes. In addition, the vxdump, vxrestore, mv_vxfs, cp_vxfs, and cpio_vxfs
commands preserve extent attributes when a file is backed up, moved, copied, or
archived.
The two basic extent attributes associated with a file are its reservation and its
fixed extent size. You can preallocate space to the file by manipulating a file's
reservation, or override the default allocation policy of the file system by setting
a fixed extent size.
Other policies determine the way these attributes are expressed during the
allocation process.
You can specify the following attribute properties:
■ The space reserved for a file must be contiguous
■ No allocations are made for a file beyond the current reservation
■ An unused reservation is released when the file is closed
■ Space is allocated, but no reservation is assigned
56 Extent attributes
■ The file size is changed to incorporate the allocated space immediately

Some of the extent attributes are persistent and become part of the on-disk
information about the file, while other attributes are temporary and are lost after
the file is closed or the system is rebooted. The persistent attributes are similar
to the file's permissions and are written in the inode for the file. When a file is
copied, moved, or archived, only the persistent attributes of the source file are
preserved in the new file.
See “Other controls” on page 57.
In general, the user will only set extent attributes for reservation. Many of the
attributes are designed for applications that are tuned to a particular pattern of
I/O or disk alignment.
See the mkfs_vxfs(1M) manual page.
See “About VxFS I/O” on page 61.
Reservation: preallocating space to a file

VxFS makes it possible to preallocate space to a file at the time of the request
rather than when data is written into the file. This space cannot be allocated to
other files in the file system. VxFS prevents any unexpected out-of-space condition
on the file system by ensuring that a file's required space will be associated with
the file before it is required.
A persistent reservation is not released when a file is truncated. The reservation
must be cleared or the file must be removed to free the reserved space.
Fixed extent size

The VxFS default allocation policy uses a variety of methods to determine how
to make an allocation to a file when a write requires additional space. The policy
attempts to balance the two goals of optimum I/O performance through large
allocations and minimal file system fragmentation through allocation from space
available in the file system that best fits the data.
Setting a fixed extent size overrides the default allocation policies for a file and
always serves as a persistent attribute. Be careful to choose an extent size
appropriate to the application when using fixed extents. An advantage of VxFS's
extent-based allocation policies is that they rarely use indirect blocks compared
to block based file systems; VxFS eliminates many instances of disk access that
stem from indirect references. However, a small extent size can eliminate this
advantage.
Files with large extents tend to be more contiguous and have better I/O
characteristics. However, the overall performance of the file system degrades
Extent attributes 57
because the unused space fragments free space by breaking large extents into
smaller pieces. By erring on the side of minimizing fragmentation for the file
system, files may become so non-contiguous that their I/O characteristics would
degrade.
Fixed extent sizes are particularly appropriate in the following situations:
■ If a file is large and contiguous, a large fixed extent size can minimize the
number of extents in the file.
Custom applications may also use fixed extent sizes for specific reasons, such as
the need to align extents to cylinder or striping boundaries on disk.
Other controls
The auxiliary controls on extent attributes determine the following conditions:
■ Whether allocations are aligned
■ Whether allocations are contiguous
■ Whether the file can be written beyond its reservation
■ Whether an unused reservation is released when the file is closed
■ Whether the reservation is a persistent attribute of the file
■ When the space reserved for a file will actually become part of the file
Alignment
Specific alignment restrictions coordinate a file's allocations with a particular
I/O pattern or disk alignment. Alignment can only be specified if a fixed extent
size has also been set. Setting alignment restrictions on allocations is best left to
well-designed applications.
See the mkfs_vxfs(1M) manual page.
See “About VxFS I/O” on page 61.
Contiguity
A reservation request can specify that its allocation remain contiguous (all one
extent). Maximum contiguity of a file optimizes its I/O characteristics.
Note: Fixed extent sizes or alignment cause a file system to return an error message
reporting insufficient space if no suitably sized (or aligned) extent is available.
This can happen even if the file system has sufficient free space and the fixed
extent size is large.
Commands related to extent attributes
Write operations beyond reservation

A reservation request can specify that no allocations can take place after a write
operation fills up the last available block in the reservation. This specification
can be used in a similar way to ulimit to prevent a file's uncontrolled growth.
Reservation trimming
A reservation request can specify that any unused reservation be released when
the file is closed. The file is not completely closed until all processes open against
the file have closed it.
Reservation persistence
A reservation request can ensure that the reservation does not become a persistent
attribute of the file. The unused reservation is discarded when the file is closed.
Including reservation in the file

A reservation request can make sure the size of the file is adjusted to include the
reservation. Normally, the space of the reservation is not included in the file until
an extending write operation requires it. A reservation that immediately changes
the file size can generate large temporary files. Unlike a ftruncate operation that
increases the size of a file, this type of reservation does not perform zeroing of
the blocks included in the file and limits this facility to users with appropriate
privileges. The data that appears in the file may have been previously contained
in another file. For users who do not have the appropriate privileges, there is a
variant request that prevents such users from viewing uniinitialized data.

The VxFS commands for manipulating extent attributes are setext and getext;
they allow the user to set up files with a given set of extent attributes or view any
attributes that are already associated with a file.
The VxFS-specific commands vxdump, vxrestore, mv_vxfs, cp_vxfs, and cpio_vxfs
preserve extent attributes when backing up, restoring, moving, or copying files.
Make sure to modify your PATH when using the VxFS versions of mv, cp, and
cpio.
Most of these commands include a command line option (-e) for maintaining
extent attributes on files. This option specifies dealing with a VxFS file that has
extent attribute information including reserved space, a fixed extent size, and
extent alignment. The extent attribute information may be lost if the destination
file system does not support extent attributes, has a different block size than the
Extent attributes 59
source file system, or lacks free extents appropriate to satisfy the extent attribute
requirements.
The -e option takes any of the following keywords as an argument:
warn Issues a warning message if extent attribute information cannot be

maintained (the default)
force Fails the copy if extent attribute information cannot be maintained
ignore Ignores extent attribute information entirely
Failure to preserve extent attributes

Whenever a file is copied, moved, or archived using commands that preserve
extent attributes, there is nevertheless the possibility of losing the attributes.
Such a failure might occur for one of the following reasons:
■ The file system receiving a copied, moved, or restored file from an archive is
not a VxFS type. Since other file system types do not support the extent
attributes of the VxFS file system, the attributes of the source file are lost
during the migration.
■ The file system receiving a copied, moved, or restored file is a VxFS type but
does not have enough free space to satisfy the extent attributes. For example,
consider a 50K file and a reservation of 1 MB. If the target file system has 500K
free, it could easily hold the file but fail to satisfy the reservation.
■ The file system receiving a copied, moved, or restored file from an archive is
a VxFS type but the different block sizes of the source and target file system
make extent attributes impossible to maintain. For example, consider a source
file system of block size 1024, a target file system of block size 4096, and a file
that has a fixed extent size of 3 blocks (3072 bytes). This fixed extent size
adapts to the source file system but cannot translate onto the target file system.
The same source and target file systems in the preceding example with a file
carrying a fixed extent size of 4 could preserve the attribute; a 4 block (4096
byte) extent on the source file system would translate into a 1 block extent on
the target.
On a system with mixed block sizes, a copy, move, or restoration operation
may or may not succeed in preserving attributes. It is recommended that the
same block size be used for all file systems on a given system.
Chapter 4
VxFS I/O Overview
■ About VxFS I/O
■ Buffered and Direct I/O
■ Cache advisories
■ Freezing and thawing a file system
■ Getting the I/O size
About VxFS I/O

VxFS processes two basic types of file system I/O:
■ Sequential
■ Random or I/O that is not sequential
For sequential I/O, VxFS employs a read-ahead policy by default when the
application is reading data. For writing, it allocates contiguous blocks if possible.
In most cases, VxFS handles I/O that is sequential through buffered I/O. VxFS
handles random or nonsequential I/O using direct I/O without buffering.
VxFS provides a set of I/O cache advisories for use when accessing files.
See the Veritas File System Programmer's Reference Guide.
See the vxfsio(7) manual page.
62 VxFS I/O Overview
Buffered and Direct I/O

VxFS responds with read-ahead for sequential read I/O. This results in buffered
I/O. The data is prefetched and retained in buffers for the application. The data
buffers are commonly referred to as VxFS buffer cache. This is the default VxFS
behavior.
On the other hand, direct I/O does not buffer the data when the I/O to the
underlying device is completed. This saves system resources like memory and
CPU usage. Direct I/O is possible only when alignment and sizing criteria are
satisfied.
See “Direct I/O requirements” on page 62.
All the supported platforms have a VxFS buffered cache. Each platform also has
either a page cache or its own buffer cache. These caches are commonly known
as the file system caches.
Direct I/O does not use these caches. The memory used for direct I/O is discarded
after the I/O is complete,
Direct I/O
Direct I/O is an unbuffered form of I/O. If the VX_DIRECT advisory is set, the user
is requesting direct data transfer between the disk and the user-supplied buffer
for reads and writes. This bypasses the kernel buffering of data, and reduces the
CPU overhead associated with I/O by eliminating the data copy between the kernel
buffer and the user's buffer. This also avoids taking up space in the buffer cache
that might be better used for something else. The direct I/O feature can provide
significant performance gains for some applications.
The direct I/O and VX_DIRECT advisories are maintained on a per-file-descriptor
basis.
Direct I/O requirements

For an I/O operation to be performed as direct I/O, it must meet certain alignment
criteria. The alignment constraints are usually determined by the disk driver, the
disk controller, and the system memory management hardware and software.
The requirements for direct I/O are as follows:
■ The starting file offset must be aligned to a 512-byte boundary.
■ The ending file offset must be aligned to a 512-byte boundary, or the length
must be a multiple of 512 bytes.
■ The memory buffer must start on an 8-byte boundary.
VxFS I/O Overview 63
Direct I/O versus synchronous I/O

Because direct I/O maintains the same data integrity as synchronous I/O, it can
be used in many applications that currently use synchronous I/O. If a direct I/O
request does not allocate storage or extend the file, the inode is not immediately
written.
Direct I/O CPU overhead

The CPU cost of direct I/O is about the same as a raw disk transfer. For sequential
I/O to very large files, using direct I/O with large transfer sizes can provide the
same speed as buffered I/O with much less CPU overhead.
If the file is being extended or storage is being allocated, direct I/O must write the
inode change before returning to the application. This eliminates some of the
performance advantages of direct I/O.
Discovered Direct I/O

Discovered Direct I/O is a file system tunable that is set using the vxtunefs
command. When the file system gets an I/O request larger than the
discovered_direct_iosz, it tries to use direct I/O on the request. For large I/O sizes,
Discovered Direct I/O can perform much better than buffered I/O.
Discovered Direct I/O behavior is similar to direct I/O and has the same alignment
constraints, except writes that allocate storage or extend the file size do not require
writing the inode changes before returning to the application.
See “Tuning I/O” on page 43..
Unbuffered I/O
If the VX_UNBUFFERED advisory is set, I/O behavior is the same as direct I/O
with the VX_DIRECT advisory set, so the alignment constraints that apply to
direct I/O also apply to unbuffered I/O. For unbuffered I/O, however, if the file is
being extended, or storage is being allocated to the file, inode changes are not
updated synchronously before the write returns to the user. The VX_UNBUFFERED
advisory is maintained on a per-file-descriptor basis.
Data synchronous I/O

If the VX_DSYNC advisory is set, the user is requesting data synchronous I/O. In
synchronous I/O, the data is written, and the inode is written with updated times
and (if necessary) an increased file size. In data synchronous I/O, the data is
transferred to disk synchronously before the write returns to the user. If the file
is not extended by the write, the times are updated in memory, and the call returns
Cache advisories
to the user. If the file is extended by the operation, the inode is written before the
write returns.
The direct I/O and VX_DSYNC advisories are maintained on a per-file-descriptor
basis.
Data synchronous I/O vs. synchronous I/O

Like direct I/O, the data synchronous I/O feature can provide significant
application performance gains. Because data synchronous I/O maintains the same
data integrity as synchronous I/O, it can be used in many applications that
currently use synchronous I/O. If the data synchronous I/O does not allocate
storage or extend the file, the inode is not immediately written. The data
synchronous I/O does not have any alignment constraints, so applications that
find it difficult to meet the alignment constraints of direct I/O should use data
synchronous I/O.
If the file is being extended or storage is allocated, data synchronous I/O must
write the inode change before returning to the application. This case eliminates
the performance advantage of data synchronous I/O.
Cache advisories
VxFS allows an application to set cache advisories for use when accessing files.
VxFS cache advisories enable applications to help monitor the buffer cache and
provide information on how better to tune the buffer cache to improve performance
gain.
The basic function of the cache advisory is to let you know whether you could
have avoided a later re-read of block X if the buffer cache had been a little larger.
Conversely, the cache advisory can also let you know that you could safely reduce
the buffer cache size without putting block X into jeopardy.
These advisories are in memory only and do not persist across reboots. Some
advisories are currently maintained on a per-file, not a per-file-descriptor, basis.
Only one set of advisories can be in effect for all accesses to the file. If two
conflicting applications set different advisories, both must use the advisories that
were last set.
All advisories are set using the VX_SETCACHE ioctl command. The current set of
advisories can be obtained with the VX_GETCACHE ioctl command.
See the vxfsio(7) manual page.
VxFS I/O Overview 65
Freezing and thawing a file system
Freezing and thawing a file system

Freezing a file system is a necessary step for obtaining a stable and consistent
image of the file system at the volume level. Consistent volume-level file system
images can be obtained and used with a file system snapshot tool. The freeze
operation flushes all buffers and pages in the file system cache that contain dirty
metadata and user data. The operation then suspends any new activity on the file
system until the file system is thawed.
The VX_FREEZE ioctl command is used to freeze a file system. Freezing a file
system temporarily blocks all I/O operations to a file system and then performs
a sync on the file system. When the VX_FREEZE ioctl is issued, all access to the
file system is blocked at the system call level. Current operations are completed
and the file system is synchronized to disk.
When the file system is frozen, any attempt to use the frozen file system, except
for a VX_THAW ioctl command, is blocked until a process executes the VX_THAW
ioctl command or the time-out on the freeze expires.
Getting the I/O size

VxFS provides the VX_GET_IOPARAMETERS ioctl to get the recommended I/O
sizes to use on a file system. This ioctl can be used by the application to make
decisions about the I/O sizes issued to VxFS for a file or file device.
See the vxtunefs(1M) and vxfsio(7) manual pages.
See “Tuning I/O” on page 43.

decisions about the I/O sizes issued to VxFS for a file or file device. For more
details on this ioctl, refer to the vxfsio(7) manual page.
For a discussion on various I/O parameters, refer to “VxFS performance: creating,
mounting, and tuning File Systems” on page 31 and the vxtunefs(1M) manual
page.

page.

page.
Chapter 5
Storage Checkpoints
■ About Storage Checkpoints
■ How a Storage Checkpoint works
■ Types of Storage Checkpoints
■ Storage Checkpoint administration
■ Space management considerations
■ Restoring a file system from a Storage Checkpoint
■ Storage Checkpoint quotas
About Storage Checkpoints

Veritas File System provides a Storage Checkpoint feature that quickly creates a
persistent image of a file system at an exact point in time. Storage Checkpoints
significantly reduce I/O overhead by identifying and maintaining only the file
system blocks that have changed since the last Storage Checkpoint or backup via
a copy-on-write technique.
See “Copy-on-write” on page 71.
Storage Checkpoints provide:
■ Persistence through reboots and crashes.
■ The ability for data to be immediately writeable by preserving the file system
metadata, the directory hierarchy, and user data.
Storage Checkpoints are actually data objects that are managed and controlled
by the file system. You can create, remove, and rename Storage Checkpoints
because they are data objects with associated names.
68 Storage Checkpoints
About Storage Checkpoints
See “How a Storage Checkpoint works” on page 69.

Unlike a disk-based mirroring technology that requires a separate storage space,
Storage Checkpoints minimize the use of disk space by using a Storage Checkpoint
within the same free space available to the file system.
After you create a Storage Checkpoint of a mounted file system, you can also
continue to create, remove, and update files on the file system without affecting
the logical image of the Storage Checkpoint. A Storage Checkpoint preserves not
only the name space (directory hierarchy) of the file system, but also the user data
as it existed at the moment the file system image was captured.
You can use a Storage checkpoints in many ways. For example, you can use them
to:
■ Create a stable image of the file system that can be backed upt backup to tape.
■ Provide a mounted, on-disk backup of the file system so that end users can
restore their own files in the event of accidental deletion. This is especially
useful in a home directory, engineering, or email environment.
■ Create a copy of an application's binaries before installing a patch to allow for
rollback in case of problems.
■ Create an on-disk backup of the file system in that can be used addition to a
traditional tape-based backup to provide faster backup and restore capabilities.
How Storage Checkpoints differ from snapshots

Storage Checkpoints differ from Veritas File System snapshots in the following
ways because they:
■ Allow write operations to the Storage Checkpoint itself.
■ Persist after a system reboot or failure.
■ Share the same pool of free space as the file system.
■ Maintain a relationship with other Storage Checkpoints by identifying changed
file blocks since the last Storage Checkpoint.
■ Have multiple, read-only Storage Checkpoints that reduce I/O operations and
required storage space because the most recent Storage Checkpoint is the only
one that accumulates updates from the primary file system.
Various backup and replication solutions can take advantage of Storage
Checkpoints. The ability of Storage Checkpoints to track the file system blocks
that have changed since the last Storage Checkpoint facilitates backup and
replication applications that only need to retrieve the changed data. Storage
Checkpoints significantly minimize data movement and may promote higher
Storage Checkpoints 69
How a Storage Checkpoint works
availability and data integrity by increasing the frequency of backup and

replication solutions.
Storage Checkpoints can be taken in environments with a large number of files
(for example, file servers with millions of files) with little adverse impact on
performance. Because the file system does not remain frozen during Storage
Checkpoint creation, applications can access the file system even while the Storage
Checkpoint is taken. However, Storage Checkpoint creation may take several
minutes to complete depending on the number of files in the file system.

The Storage Checkpoint facility freezes the mounted file system (known as the
primary fileset), initializes the Storage Checkpoint, and thaws the file system.
Specifically, the file system is first brought to a stable state where all of its data
is written to disk, and the freezing process momentarily blocks all I/O operations
to the file system. A Storage Checkpoint is then created without any actual data;
the Storage Checkpoint instead points to the block map of the primary fileset.
The thawing process that follows restarts I/O operations to the file system.
You can create a Storage Checkpoint on a single file system or a list of file systems.
A Storage Checkpoint of multiple file systems simultaneously freezes the file
systems, creates a Storage Checkpoint on all of the file systems, and thaws the
file systems. As a result, the Storage Checkpoints for multiple file systems have
the same creation timestamp. The Storage Checkpoint facility guarantees that
multiple file system Storage Checkpoints are created on all or none of the specified
file systems, unless there is a system crash while the operation is in progress.
Note: The calling application is responsible for cleaning up Storage Checkpoints

after a system crash.
A Storage Checkpoint of the primary fileset initially contains a pointer to the file
system block map rather than to any actual data. The block map points to the data
on the primary fileset.
Figure 5-1 shows the file system /database and its Storage Checkpoint.
The Storage Checkpoint is logically identical to the primary fileset when the
Storage Checkpoint is created, but it does not contain any actual data blocks.
Figure 5-1 Primary fileset and its Storage Checkpoint
Primary fileset Storage Checkpoint
/database /database
emp.dbf jun.dbf emp.dbf jun.dbf
In Figure 5-2, a square represents each block of the file system. This figure shows
a Storage Checkpoint containing pointers to the primary fileset at the time the
Storage Checkpoint is taken, as in Figure 5-1.
Figure 5-2 Initializing a Storage Checkpoint
E
The Storage Checkpoint presents the exact image of the file system by finding
the data from the primary fileset. As the primary fileset is updated, the original
data is copied to the Storage Checkpoint before the new data is written. When a
write operation changes a specific data block in the primary fileset, the old data
is first read and copied to the Storage Checkpoint before the primary fileset is
updated. Subsequent writes to the specified data block on the primary fileset do
not result in additional updates to the Storage Checkpoint because the old data
needs to be saved only once. As blocks in the primary fileset continue to change,
the Storage Checkpoint accumulates the original data blocks.
Copy-on-write
In Figure 5-3, the third block originally containing C is updated.
Before the block is updated with new data, the original data is copied to the Storage
Checkpoint. This is called the copy-on-write technique, which allows the Storage
Checkpoint to preserve the image of the primary fileset when the Storage
Checkpoint is taken.
Every update or write operation does not necessarily result in the process of
copying data to the Storage Checkpoint. In this example, subsequent updates to
this block, now containing C', are not copied to the Storage Checkpoint because
the original image of the block containing C is already saved.
Types of Storage Checkpoints
Figure 5-3 Updates to the primary fileset
C’ C

You can create the following types of Storage Checkpoints:
■ Data Storage Checkpoints
■ nodata Storage Checkpoints
■ Removable Storage Checkpoints
■ Non-mountable Storage Checkpoints
Data Storage Checkpoints

A data Storage Checkpoint is a complete image of the file system at the time the
Storage Checkpoint is created. This type of Storage Checkpoint contains the file
system metadata and file data blocks. You can mount, access, and write to a data
Storage Checkpoint just as you would to a file system. Data Storage Checkpoints
are useful for backup applications that require a consistent and stable image of
an active file system. Data Storage Checkpoints introduce some overhead to the
system and to the application performing the write operation. For best results,
limit the life of data Storage Checkpoints to minimize the impact on system
resources.
See “Showing the difference between a data and a nodata Storage Checkpoint”
on page 79.
nodata Storage Checkpoints

A nodata Storage Checkpoint only contains file system metadata—no file data
blocks. As the original file system changes, the nodata Storage Checkpoint records
the location of every changed block. Nodata Storage Checkpoints use minimal
system resources and have little impact on the performance of the file system
because the data itself does not have to be copied.
In Figure 5-4, the first block originally containing A is updated.
The original data is not copied to the storage checkpoint, but the changed block
is marked in the Storage Checkpoint. The marker indicates which data has changed.
Figure 5-4 Updates to a nodata clone
A’
See “Showing the difference between a data and a nodata Storage Checkpoint”
on page 79.
Storage Checkpoint administration
Removable Storage Checkpoints

A removable Storage Checkpoint can “self-destruct” under certain conditions
when the file system runs out of space.
See “Space management considerations” on page 86.
After encountering certain out-of-space (ENOSPC) conditions, the kernel removes
Storage Checkpoints to free up space for the application to continue running on
the file system. In almost all situations, you should create Storage Checkpoints
with the removable attribute.
Non-mountable Storage Checkpoints

You can create Storage Checkpoints that cannot be mounted by using the
fsckptadm set nomount command.
See the fsckptadm(1M) manual page.
Use this type of Storage Checkpoint as a security feature which prevents other
applications from accessing the Storage Checkpoint and modifying it.

Storage Checkpoint administrative operations require the fsckptadm utility.
You can use the fsckptadm utility to create and remove Storage Checkpoints,
change attributes, and ascertain statistical data. Every Storage Checkpoint has
an associated name, which allows you to manage Storage Checkpoints; this name
is limited to 127 characters and cannot contain a colon (:).
Storage Checkpoints require some space for metadata on the volume or set of
volumes specified by the file system allocation policy or Storage Checkpoint
allocation policy. The fsckptadm utility displays an error if the volume or set of
volumes does not have enough free space to contain the metadata. You can roughly
approximate the amount of space required by the metadata using a method that
depends on the disk layout version of the file system.
For disk layout Version 5 or prior, multiply the number of inodes (# of inodes) by
the inode size (inosize) in bytes, and add 1 or 2 megabytes to get the approximate
amount of space required. You can determine the number of inodes with the
fsckptadm utility, and the inode size with the mkfs command:
# fsckptadm -v info '' /mnt0

UNNAMED:
ctime = Thu 3 Mar 2005 7:00:17 PM PST

mtime = Thu 3 Mar 2005 7:00:17 PM PST
flags = largefiles, mounted
# of inodes = 23872
# of blocks = 27867
.
.
.
# of overlay bmaps = 0
# mkfs -m /mnt0
# mkfs -F vxfs -o \
bsize=1024,version=7,inosize=256,logsize=65536,\
largefiles /mnt0
In this example, the approximate amount of space required by the metadata is 7

or 8 megabytes (23,872 x 256 bytes, plus 1 or 2 megabytes).
For disk layout Version 6 or 7, multiply the number of inodes by 1 byte, and add
1 or 2 megabytes to get the approximate amount of space required. You can
determine the number of inodes with the fsckptadm utility as above. Using the
output from the example for disk layout Version 5, the approximate amount of
space required by the metadata is just over one or two megabytes (23,872 x 1 byte,
plus 1 or 2 megabytes).
Use the fsvoladm command to determine if the volume set has enough free space.
See the fsvoladm(1M) manual page.
# fsvoladm list /mnt0

devid size used avail name
0 20971520 8497658 12473862 mnt1
1 20971520 6328993 14642527 mnt2
2 20971520 4458462 16513058 mnt3
Creating a Storage Checkpoint

The following example shows the creation of a nodata Storage Checkpoint named
thu_7pm on /mnt0 and lists all Storage Checkpoints of the /mnt0 file system:
# fsckptadm -n create thu_7pm /mnt0

# fsckptadm list /mnt0
/mnt0
thu_7pm:
flags = nodata, largefiles
See “Space management considerations” on page 86.

The following example shows the creation of a removable Storage Checkpoint
named thu_8pm on /mnt0 and lists all Storage Checkpoints of the /mnt0 file system:
# fsckptadm -r create thu_8pm /mnt0

/mnt0
thu_8pm:
flags = largefiles, removable
thu_7pm:
Removing a Storage Checkpoint

You can delete a Storage Checkpoint by specifying the remove keyword of the
fsckptadm command. Specifically, you can use either the synchronous or
asynchronous method of removing a Storage Checkpoint; the asynchronous
method is the default method. The synchronous method entirely removes the
Storage Checkpoint and returns all of the blocks to the file system before
completing the fsckptadm operation. The asynchronous method simply marks
the Storage Checkpoint for removal and causes fsckptadm to return immediately.
At a later time, an independent kernel thread completes the removal operation
and releases the space used by the Storage Checkpoint.
In this example, /mnt0 is a mounted VxFS file system with a Version 4 disk layout.
This example shows the asynchronous removal of the Storage Checkpoint named
thu_8pm and synchronous removal of the Storage Checkpoint named thu_7pm.
This example also lists all the Storage Checkpoints remaining on the /mnt0 file
system after the specified Storage Checkpoint is removed:
# fsckptadm remove thu_8pm /mnt0

/mnt0
thu_7pm:
# fsckptadm -s remove thu_7pm /mnt0
/mnt0
Accessing a Storage Checkpoint

You can mount Storage Checkpoints using the mount command with the mount
option –o ckpt=ckpt_name.
Observe the following rules when mounting Storage Checkpoints:
■ Storage Checkpoints are mounted as read-only Storage Checkpoints by default.
If you need to write to a Storage Checkpoint, mount it using the -o rw option.
■ If a Storage Checkpoint is originally mounted as a read-only Storage
Checkpoint, you can remount it as a writable Storage Checkpoint using the -o
remount option.
■ To mount a Storage Checkpoint of a file system, first mount the file system
itself.
■ To unmount a file system, first unmount all of its Storage Checkpoints.
Warning: If you create a Storage Checkpoint for backup purposes, do not mount
it as a writable Storage Checkpoint. You will lose the point-in-time image if
you accidently write to the Storage Checkpoint.
A Storage Checkpoint is mounted on a special pseudo device. This pseudo device

does not exist in the system name space; the device is internally created by the
system and used while the Storage Checkpoint is mounted. The pseudo device is
removed after you unmount the Storage Checkpoint. A pseudo device name is
formed by appending the Storage Checkpoint name to the file system device name
using the colon character (:) as the separator.
For example, if a Storage Checkpoint named may_23 belongs to the file system
residing on the special device /dev/vx/dsk/fsvol/vol1, the Storage Checkpoint
pseudo device name is:
/dev/vx/dsk/fsvol/vol1:may_23
■ To mount the Storage Checkpoint named may_23 as a read-only (default) Storage

Checkpoint on directory /fsvol_may_23, type:
# mount -F vxfs -o ckpt=may_23 /dev/vx/dsk/fsvol/vol1:may_23 \

/fsvol_may_23
Note: The vol1 file system must already be mounted before the Storage
Checkpoint can be mounted.
■ To remount the Storage Checkpoint named may_23 as a writable Storage

Checkpoint, type:
# mount -F vxfs -o ckpt=may_23,remount,rw \

/dev/vx/dsk/fsvol/vol1:may_23 /fsvol_may_23
■ To mount this Storage Checkpoint automatically when the system starts up,
put the following entries in the /etc/vfstab file:
e c device
i v e d #to fsck mount point FS fsck mount at mount
o t type pass boot options
t n u om
/ xv/dev/vx/rdsk/fsvol/vol1
/ v ed / /fsvol vxfs 1 yes —
/ k s d
/ l ov s f
1 l o v
/ xv—/ v ed / /fsvol_may_23 vxfs 0 yes ckpt=

/ l ovs f /ksd may_23
: 1 l ov
3 2 _ y am
You can unmount Storage Checkpoints using the umount command.

See the umount_vxfs(1M) manual page.
Storage Checkpoints can be unmounted by the mount point or pseudo device
name:
# umount /fsvol_may_23
# umount /dev/vx/dsk/fsvol/vol1:may_23
/dev/vx/dsk/fsvol/vol1 /fsvol vxfs defaults 0 2
/dev/vx/dsk/fsvol/vol1:may_23 /fsvol_may_23 vxfs clone=may_23 0 0
Note: You do not need to run the fsck utility on Storage Checkpoint pseudo devices
because pseudo devices are part of the actual file system.
Converting a data Storage Checkpoint to a nodata Storage Checkpoint

A nodata Storage Checkpoint does not contain actual file data. Instead, this type
of Storage Checkpoint contains a collection of markers indicating the location of
all the changed blocks since the Storage Checkpoint was created.
See “Types of Storage Checkpoints” on page 72.

You can use either the synchronous or asynchronous method to convert a data
Storage Checkpoint to a nodata Storage Checkpoint; the asynchronous method
is the default method. In a synchronous conversion, fsckptadm waits for all files
to undergo the conversion process to “nodata” status before completing the
operation. In an asynchronous conversion, fsckptadm returns immediately and
marks the Storage Checkpoint as a nodata Storage Checkpoint even though the
Storage Checkpoint's data blocks are not immediately returned to the pool of free
blocks in the file system. The Storage Checkpoint deallocates all of its file data
blocks in the background and eventually returns them to the pool of free blocks
in the file system.
If all of the older Storage Checkpoints in a file system are nodata Storage
Checkpoints, use the synchronous method to convert a data Storage Checkpoint
to a nodata Storage Checkpoint. If an older data Storage Checkpoint exists in the
file system, use the asynchronous method to mark the Storage Checkpoint you
want to convert for a delayed conversion. In this case, the actual conversion will
continue to be delayed until the Storage Checkpoint becomes the oldest Storage
Checkpoint in the file system, or all of the older Storage Checkpoints have been
converted to nodata Storage Checkpoints.
Note: You cannot convert a nodata Storage Checkpoint to a data Storage

Checkpoint because a nodata Storage Checkpoint only keeps track of the location
of block changes and does not save the content of file data blocks.
Showing the difference between a data and a nodata Storage

Checkpoint
The following example shows the difference between data Storage Checkpoints
and nodata Storage Checkpoints.
Note: A nodata Storage Checkpoint does not contain actual file data.
See “Converting a data Storage Checkpoint to a nodata Storage Checkpoint”

on page 78.
To show the difference between Storage Checkpoints

1 Create a file system and mount it on /mnt0, as in the following example:
# mkfs -F vxfs /dev/vx/rdsk/dg1/test0
version 7 layout
134217728 sectors, 67108864 blocks of size 1024, log \
size 65536 blocks, largefiles supported

# mkfs -F /dev/vx/rdsk/dg1/test0 /mnt0
2 Create a small file with a known content, as in the following example.
# echo "hello, world" > /mnt0/file
3 Create a Storage Checkpoint and mount it on /mnt0@5_30pm, as in the

following example:
# fsckptadm create ckpt@5_30pm /mnt0

# mkdir /mnt0@5_30pm
# mount -F vxfs -o ckpt=ckpt@5_30pm \
/dev/vx/dsk/dg1/test0:ckpt@5_30pm /mnt0@5_30pm
4 Examine the content of the original file and the Storage Checkpoint file:
# cat /mnt0/file
hello, world
# cat /mnt0@5_30pm/file
hello, world
5 Change the content of the original file:
# echo "goodbye" > /mnt0/file
6 Examine the content of the original file and the Storage Checkpoint file. The
original file contains the latest data while the Storage Checkpoint file still
contains the data at the time of the Storage Checkpoint creation:
# cat /mnt0/file
goodbye
hello, world
7 Unmount the Storage Checkpoint, convert the Storage Checkpoint to a nodata

Storage Checkpoint, and mount the Storage Checkpoint again.
# umount /mnt0@5_30pm
# fsckptadm -s set nodata ckpt@5_30pm /mnt0
# mount -F vxfs -o ckpt=ckpt@5_30pm \
/dev/vx/dsk/dg1/test0:ckpt@5_30pm /mnt0@5_30pm
8 Examine the content of both files. The original file must contain the latest
data:
# cat /mnt0/file
goodbye
You can traverse and read the directories of the nodata Storage Checkpoint;
however, the files contain no data, only markers to indicate which block of
the file has been changed since the Storage Checkpoint was created:
# ls -l /mnt0@5_30pm/file
-rw-r--r-- 1 root other 13 Jul 13 17:13 \
mnt0@5_30pm/file
cat: /mnt0@5_30pm/file: I/O error
cat: read error: No such file or directory
cat: /mnt0@5_30pm/file: Input/output error
cat: input error on /mnt0@5_30pm/file: I/O error
cat: /mnt0@5_30pm/file: I/O error
Converting multiple Storage Checkpoints

You can convert Storage Checkpoints to nodata Storage Checkpoints, when dealing
with older Storage Checkpoints on the same file system.
To convert multiple Storage Checkpoints
1 Create a file system and mount it on /mnt0:
# mkfs -F vxfs /dev/vx/rdsk/dg1/test0

version 7 layout
13417728 sectors, 67108864 blocks of size 1024, log \
size 65536 blocks largefiles supported
# mount -F vxfs /dev/vx/dsk/dg1/test0 /mnt0
2 Create four data Storage Checkpoints on this file system, note the order of
creation, and list them:
# fsckptadm create oldest /mnt0

# fsckptadm create older /mnt0
# fsckptadm create old /mnt0
# fsckptadm create latest /mnt0
/mnt0
latest:
ctime = Mon 26 Jul 11:56:55 2004
mtime = Mon 26 Jul 11:56:55 2004
flags = largefiles
old:
ctime = Mon 26 Jul 11:56:51 2004
mtime = Mon 26 Jul 11:56:51 2004
flags = largefiles
older:
ctime = Mon 26 Jul 11:56:46 2004
mtime = Mon 26 Jul 11:56:46 2004
flags = largefiles
oldest:
ctime = Mon 26 Jul 11:56:41 2004
mtime = Mon 26 Jul 11:56:41 2004
flags = largefiles
3 Try to convert synchronously the “latest” Storage Checkpoint to a nodata

Storage Checkpoint. The attempt will fail because the Storage Checkpoints
older than the “latest” Storage Checkpoint are data Storage Checkpoints,
namely the Storage Checkpoints “old”, “older”, and “oldest”:
# fsckptadm -s set nodata latest /mnt0

UX:vxfs fsckptadm: ERROR: V-3-24632: Storage Checkpoint
set failed on latest. File exists (17)
4 You can instead convert the “latest” Storage Checkpoint to a nodata Storage
Checkpoint in a delayed or asynchronous manner.
# fsckptadm set nodata latest /mnt0
5 List the Storage Checkpoints, as in the following example. You will see that
the “latest” Storage Checkpoint is marked for conversion in the future.

/mnt0
latest:
ctime = Mon 26 Jul 11:56:55 2004
mtime = Mon 26 Jul 11:56:55
flags = nodata, largefiles, delayed
old:
ctime = Mon 26 Jul 11:56:51 2004
mtime = Mon 26 Jul 11:56:51 2004
flags = largefiles
older:
ctime = Mon 26 Jul 11:56:46 2004
mtime = Mon 26 Jul 11:56:46 2004
flags = largefiles
oldest:
ctime = Mon 26 Jul 11:56:41 2004
mtime = Mon 26 Jul 11:56:41 2004
flags = largefiles
Creating a delayed nodata Storage Checkpoint

You can combine the three previous steps and create the “latest” Storage
Checkpoint as a nodata Storage Checkpoint. The creation process will detect the
presence of the older data Storage Checkpoints and create the “latest” Storage
Checkpoint as a delayed nodata Storage Checkpoint.
To create a delayed nodata Storage Checkpoint

1 Remove the “latest” Storage Checkpoint.
# fsckptadm remove latest /mnt0

/mnt0
old:
ctime = Mon 26 Jul 11:56:51 2004
mtime = Mon 26 Jul 11:56:51 2004
flags = largefiles
older:
ctime = Mon 26 Jul 11:56:46 2004
mtime = Mon 26 Jul 11:56:46 2004
flags = largefiles
oldest:
ctime = Mon 26 Jul 11:56:41 2004
mtime = Mon 26 Jul 11:56:41 2004
flags = largefiles
2 Recreate the latest Storage Checkpoint as a nodata Storage Checkpoint.
# fsckptadm -n create latest /mnt0

/mnt0
latest:
ctime = Mon 26 Jul 12:06:42 2004
mtime = Mon 26 Jul 12:06:42 2004
old:
ctime = Mon 26 Jul 11:56:51 2004
mtime = Mon 26 Jul 11:56:51 2004
flags = largefiles
older:
ctime = Mon 26 Jul 11:56:46 2004
mtime = Mon 26 Jul 11:56:46 2004
flags = largefiles
oldest:
ctime = Mon 26 Jul 11:56:41 2004
mtime = Mon 26 Jul 11:56:41 2004
flags = largefiles
3 Convert the “oldest” Storage Checkpoint to a nodata Storage Checkpoint

because no older Storage Checkpoints exist that contain data in the file system.
Note: This step can be done synchronously.
# fsckptadm -s set nodata oldest /mnt0

/mnt0
latest:
ctime = Mon 26 Jul 12:06:42 2004
mtime = Mon 26 Jul 12:06:42 2004
old:
ctime = Mon 26 Jul 11:56:51 2004
mtime = Mon 26 Jul 11:56:51 2004
flags = largefiles
older:
ctime = Mon 26 Jul 11:56:46 2004
mtime = Mon 26 Jul 11:56:46 2004
flags = largefiles
oldest:
ctime = Mon 26 Jul 11:56:41 2004
mtime = Mon 26 Jul 11:56:41 2004
Space management considerations
4 Remove the “older” and “old” Storage Checkpoints.
# fsckptadm remove older /mnt0

# fsckptadm remove old /mnt0
/mnt0
latest:
ctime = Mon 26 Jul 12:06:42 2004
mtime = Mon 26 Jul 12:06:42 2004
oldest:
ctime = Mon 26 Jul 11:56:41 2004
mtime = Mon 26 Jul 11:56:41 2004
Note: After you remove the “older” and “old” Storage Checkpoints, the “latest”
Storage Checkpoint is automatically converted to a nodata Storage Checkpoint
because the only remaining older Storage Checkpoint (“oldest”) is already a
nodata Storage Checkpoint:
Space management considerations

Several operations, such as removing or overwriting a file, can fail when a file
system containing Storage Checkpoints runs out of space. If the system cannot
allocate sufficient space, the operation will fail.
Database applications usually preallocate storage for their files and may not
expect a write operation to fail. If a file system runs out of space, the kernel
automatically removes Storage Checkpoints and attempts to complete the write
operation after sufficient space becomes available. The kernel removes Storage
Checkpoints to prevent commands, such as rm. from failing under an out-of-space
(ENOSPC) condition.
See the rm(1m) manual page.
When the kernel automatically removes the Storage Checkpoints, it applies the
following policies:
■ Remove as few Storage Checkpoints as possible to complete the operation.
■ Never select a non-removable Storage Checkpoint.
■ Select a nodata Storage Checkpoint only when data Storage Checkpoints no
longer exist.
Restoring a file system from a Storage Checkpoint
■ Remove the oldest Storage Checkpoint first.

Mountable data Storage Checkpoints on a consistent and undamaged file system
can be used by backup and restore applications to restore either individual files
or an entire file system. Restoration from Storage Checkpoints can also help
recover incorrectly modified files, but typically cannot recover from hardware
damage or other file system integrity problems.
Note: For hardware or other integrity problems, Storage Checkpoints must be

supplemented by backups from other media.
Files can be restored by copying the entire file from a mounted Storage Checkpoint
back to the primary fileset. To restore an entire file system, you can designate a
mountable data Storage Checkpoint as the primary fileset using the
fsckpt_restore command.
See the fsckpt_restore(1M) manual page.

When using the fsckpt_restore command to restore a file system from a Storage
Checkpoint, all changes made to that file system after that Storage Checkpoint's
creation date are permanently lost. The only Storage Checkpoints and data
preserved are those that were created at the same time, or before, the selected
Storage Checkpoint's creation. The file system cannot be mounted when
fsckpt_restore is invoked.
Note: Files can be restored very efficiently by applications using the

fsckpt_fbmap(3) library function to restore only modified portions of a files data.
Restoring a file from a Storage Checkpoint

The following example restores a file, MyFile.txt, which resides in your home
directory, from the Storage Checkpoint “CKPT1” to the device /dev/vx/dsk/vol-01.
The mount point for the device is /home.
To restore a file from a Storage Checkpoint

1 Create the Storage Checkpoint CKPT1 of /home.
$ fckptadm create CKPT1 /home
2 Mount Storage Checkpoint CKPT1 on the directory /home/checkpoints/mar_4.
$ mount -FFFtV vxfs -o ckpt=CKPT1 /dev/vx/dsk/dg1/vol- \

01:CKPT1/home/checkpoints/mar_4
3 Delete the file MyFile.txt from your home directory.
$ cd /home/users/me
$ rm MyFile.txt
4 Go to the /home/checkpoints/mar_4/users/me directory, which contains the

image of your home directory.
$ cd /home/checkpoints/mar_4/users/me
$ ls -l
-rw-r--r-- 1 me staff 14910 Mar 4 17:09 MyFile.txt
5 Copy the file MyFile.txt to your home directory.
$ cp MyFile.txt /home/users/me
$ cd /home/users/me
$ ls -l
-rw-r--r-- 1 me staff 14910 Mar 4 18:21 MyFile.txt

The following example restores a file system from the Storage Checkpoint “CKPT3.”
The filesets listed before the restoration show an unnamed root fileset and six
Storage Checkpoints.
U
C C C C C C
N
K K K K K K
A
P P P P P P
M
T T T T T T
E
6 5 4 3 2 1
D
To restore a file system from a Storage Checkpoint

1 Run the fsckpt_restore command:
# fsckpt_restore -l /dev/vx/dsk/dg1/vol2
/dev/vx/dsk/dg1/vol2:
UNNAMED:
ctime = Thu 08 May 2004 06:28:26 PM PST
mtime = Thu 08 May 2004 06:28:26 PM PST
flags = largefiles, file system root
CKPT6:
flags = largefiles
CKPT5:
flags = largefiles, nomount
CKPT4:
flags = largefiles
CKPT3:
flags = largefiles
CKPT2:
flags = largefiles
CKPT1:
2 In this example, select the Storage Checkpoint “CKPT3” as the new root fileset:
Select Storage Checkpoint for restore operation

or <Control/D> (EOF) to exit
or <Return> to list Storage Checkpoints: CKPT3
CKPT3:
flags = largefiles
UX:vxfs fsckpt_restore: WARNING: V-3-24640: Any file system
changes or Storage Checkpoints made after
Thu 08 May 2004 06:28:31 PM PST will be lost.
3 Type y to restore the file system from CKPT3:
Restore the file system from Storage Checkpoint CKPT3 ?

(ynq) y
(Yes)
UX:vxfs fsckpt_restore: INFO: V-3-23760: File system
restored from CKPT3
If the filesets are listed at this point, it shows that the former UNNAMED root
fileset and CKPT6, CKPT5, and CKPT4 were removed, and that CKPT3 is now
the primary fileset. CKPT3 is now the fileset that will be mounted by default.
C C C
K K K
P P P
T T T
3 2 1
4 Run the fsckpt_restore command:
# fsckpt_restore -l /dev/vx/dsk/dg1/vol2
/dev/vx/dsk/dg1/vol2:
CKPT3:
flags = largefiles, file system root
CKPT2:
flags = largefiles
CKPT1:
Select Storage Checkpoint for restore operation
or <Control/D> (EOF) to exit
or <Return> to list Storage Checkpoints:
Storage Checkpoint quotas
Storage Checkpoint quotas

VxFS provides options to the fsckptadm command interface to administer Storage
Checkpoint quotas. Storage Checkpoint quotas set the following limits on the
number of blocks used by all Storage Checkpoints of a primary file set:
hard limit An absolute limit that cannot be exceeded. If a hard limit is exceeded,
all further allocations on any of the Storage Checkpoints fail, but
existing Storage Checkpoints are preserved.
soft limit Must be lower than the hard limit. If a soft limit is exceeded, no new
Storage Checkpoints can be created. The number of blocks used must
return below the soft limit before more Storage Checkpoints can be
created. An alert and console message are generated.
In case of a hard limit violation, various solutions are possible, enacted by

specifying or not specifying the -f option for the fsckptadm utility.
Specifying or not specifying the -f option has the following effects:
■ If the -f option is not specified, one or many removable Storage Checkpoints
are deleted to make space for the operation to succeed. This is the default
solution.
■ If the -f option is specified, all further allocations on any of the Storage
Checkpoints fail, but existing Storage Checkpoints are preserved.
Note: Sometimes if a file is removed while it is opened by another process, the

removal process is deferred until the last close. Because the removal of a file
may trigger pushing data to a “downstream” Storage Checkpoint (that is, the
next older Storage Checkpoint), a fileset hard limit quota violation may occur.
In this scenario, the hard limit is relaxed to prevent an inode from being marked
bad. This is also true for some asynchronous inode operations.
Chapter 6
Online backup using file
system snapshots
■ About snapshot file systems
■ Snapshot file system backups
■ Creating a snapshot file system
■ Backup examples
■ Snapshot file system performance
■ Differences between snapshots and Storage Checkpoints
■ About snapshot file system disk structure
■ How a snapshot file system works
About snapshot file systems

A snapshot file system is an exact image of a VxFS file system, referred to as the
snapped file system, that provides a mechanism for making backups. The snapshot
is a consistent view of the file system “snapped” at the point in time the snapshot
is made. You can select files to back up from the snapshot using a standard utility
such as cpio or cp, or back up the entire file system image using the vxdump or
fscat utilities.
You use the mount command to create a snapshot file system; the mkfs command
is not required. A snapshot file system is always read-only. A snapshot file system
exists only as long as it and the snapped file system are mounted and ceases to
exist when unmounted. A snapped file system cannot be unmounted until all of
94 Online backup using file system snapshots
Snapshot file system backups
its snapshots are unmounted. Although it is possible to have multiple snapshots

of a file system made at different times, it is not possible to make a snapshot of a
snapshot.
Note: A snapshot file system ceases to exist when unmounted. If mounted again,
it is actually a fresh snapshot of the snapped file system. A snapshot file system
must be unmounted before its dependent snapped file system can be unmounted.
Neither the fuser command nor the mount command will indicate that a snapped
file system cannot be unmounted because a snapshot of it exists.
On cluster file systems, snapshots can be created on any node in the cluster, and
backup operations can be performed from that node. The snapshot of a cluster
file system is accessible only on the node where it is created, that is, the snapshot
file system itself cannot be cluster mounted.
See the Veritas Storage Foundation Cluster File System Administrator's Guide.
Snapshot file system backups

After a snapshot file system is created, the snapshot maintains a consistent backup
of data in the snapped file system.
Backup programs, such as cpio, that back up a standard file system tree can be
used without modification on a snapshot file system because the snapshot presents
the same data as the snapped file system. Backup programs, such as vxdump, that
access the disk structures of a file system require some modifications to handle
a snapshot file system.
VxFS utilities recognize snapshot file systems and modify their behavior so that
they operate the same way on snapshots as they do on standard file systems.
Other backup programs that typically read the raw disk image cannot work on
snapshots without altering the backup procedure.
These other backup programs can use the fscat command to obtain a raw image
of the entire file system that is identical to an image obtainable by running a dd
command on the disk device containing the snapped file system at the exact
moment the snapshot was created. The snapread ioctl takes arguments similar
to those of the read system call and returns the same results that are obtainable
by performing a read on the disk device containing the snapped file system at the
exact time the snapshot was created. In both cases, however, the snapshot file
system provides a consistent image of the snapped file system with all activity
complete—it is an instantaneous read of the entire file system. This is much
different than the results that would be obtained by a dd or read command on the
disk device of an active file system.
Online backup using file system snapshots 95
Creating a snapshot file system
Creating a snapshot file system

You create a snapshot file system by using the -o snapof= option of the mount
command. The -o snapsize= option may also be required if the device you are
mounting does not identify the device size in its disk label, or if you want a size
smaller than the entire device.
You must make the snapshot file system large enough to hold any blocks on the
snapped file system that may be written to while the snapshot file system exists.
If a snapshot runs out of blocks to hold copied data, the snapshot is disabled and
further attempts to access the snapshot file system fail.
During periods of low activity (such as nights and weekends), a snapshot typically
requires about two to six percent of the blocks of the snapped file system. During
a period of high activity, the snapshot of a typical file system may require 15
percent of the blocks of the snapped file system. Most file systems do not turn
over 15 percent of data in a single day. These approximate percentages tend to
be lower for larger file systems and higher for smaller file systems. You can allocate
blocks to a snapshot based on characteristics such as file system usage and
duration of backups.
Warning: Any existing data on the device used for the snapshot is overwritten.
To create a snapshot file system

◆ Mount the file system with the -o snapof= option:
# mount -F vxfs -o snapof=special,snapsize=snapshot_size \

snapshot_special snapshot_mount_point




Backup examples
Backup examples
In the following examples, the vxdump utility is used to ascertain whether
/dev/vx/dsk/fsvol/vol1 is a snapshot mounted as /backup/home and does the
appropriate work to get the snapshot data through the mount point.
These are typical examples of making a backup of a 300,000 block file system
named /home using a snapshot file system on /dev/vx/dsk/fsvol/vol1 with a
snapshot mount point of /backup/home.
To create a backup using a snapshop file system
1 To back up files changed within the last week using cpio:
# mount -F vxfs -o snapof=/home,snapsize=100000 \

/dev/vx/dsk/fsvol/vol1 /backup/home
# cd /backup
# find home -ctime -7 -depth -print | cpio -oc > \
/dev/rmt/c0s0
# umount /backup/home
2 To do a level 3 backup of /dev/vx/dsk/fsvol/vol1 and collect those files

that have changed in the current directory:
# vxdump 3f - /dev/vx/dsk/fsvol/vol1 | vxrestore -xf -
3 To do a full backup of /home, which exists on disk /dev/vx/dsk/fsvol/vol1,

and use dd to control blocking of output onto tape device using vxdump:
# mount -F vxfs -o snapof=/home,snapsize=100000 \

# vxdump f - /dev/vx/dsk/fsvol/vol1 | dd bs=128k > \
/dev/rmt/c0s0
Snapshot file system performance
To create a backup using a snapshop file system

1 To back up files changed within the last week using cpio:
# mount -F vxfs –o snapof=/home,snapsize=100000 \

# cd /backup
# find home –ctime –7 –depth –print | cpio –oc > \
/dev/rmt/c0s0
# umount /backup/home
2 To do a level 3 backup of /dev/vx/dsk/fsvol/vol1 and collect those files that

have changed in the current directory:
# vxdump 3f – /dev/vx/dsk/fsvol/vol1 | vxrestore –xf –
3 To do a full backup of /home, which exists on disk /dev/vx/dsk/fsvol/vol1,

and use dd to control blocking of output onto tape device using vxdump:
# mount –F vxfs –o snapof=/home,snapsize=100000 \

# vxdump f – /dev/vx/dsk/fsvol/vol1 | dd bs=128k > \
/dev/rmt/c0s0
Snapshot file system performance

Snapshot file systems maximize the performance of the snapshot at the expense
of writes to the snapped file system. Reads from a snapshot file system typically
perform at nearly the throughput rates of reads from a standard VxFS file system.
The performance of reads from the snapped file system are generally not affected.
However, writes to the snapped file system, typically average two to three times
as long as without a snapshot. This is because the initial write to a data block
requires reading the old data, writing the data to the snapshot, and then writing
the new data to the snapped file system. If there are multiple snapshots of the
same snapped file system, writes are even slower. Only the initial write to a block
experiences this delay, so operations such as writes to the intent log or inode
updates proceed at normal speed after the initial write.
Differences between snapshots and Storage Checkpoints
Reads from the snapshot file system are impacted if the snapped file system is
busy because the snapshot reads are slowed by the disk I/O associated with the
snapped file system.
The overall impact of the snapshot is dependent on the read to write ratio of an
application and the mixing of the I/O operations. For example, a database
application running an online transaction processing (OLTP) workload on a
snapped file system was measured at about 15 to 20 percent slower than a file
system that was not snapped.
Differences between snapshots and Storage

Checkpoints
While snapshots and Storage Checkpoints both create a point-in-time image of a
file system and only the changed data blocks are updated, there are significant
differences between the two technologies:
Table 6-1 Differences between snapshots and Storage Checkpoints
Snapshots Storage Checkpoints
Require a separate device for storage Reside on the same device as the original file
system
Are read-only Can be read-only or read-write
Are transient Are persistent
Cease to exist after being unmounted Can exist and be mounted on their own
Track changed blocks on the file system level Track changed blocks on each file in the file
system
Storage Checkpoints also serve as the enabling technology for two other Veritas
features: Block-Level Incremental Backups and Storage Rollback, which are used
extensively for backing up databases.
See “About Storage Checkpoints” on page 67.
About snapshot file system disk structure

A snapshot file system consists of:
■ A super-block
■ A bitmap
How a snapshot file system works
■ A blockmap
■ Data blocks copied from the snapped file system
The following figure shows the disk structure of a snapshot file system.
Figure 6-1 The Snapshot Disk Structure
super-block
bitmap
blockmap
data block
The super-block is similar to the super-block of a standard VxFS file system, but
the magic number is different and many of the fields are not applicable.
The bitmap contains one bit for every block on the snapped file system. Initially,
all bitmap entries are zero. A set bit indicates that the appropriate block was
copied from the snapped file system to the snapshot. In this case, the appropriate
position in the blockmap references the copied block.
The blockmap contains one entry for each block on the snapped file system.
Initially, all entries are zero. When a block is copied from the snapped file system
to the snapshot, the appropriate entry in the blockmap is changed to contain the
block number on the snapshot file system that holds the data from the snapped
file system.
The data blocks are filled by data copied from the snapped file system, starting
from the beginning of the data block area.

A snapshot file system is created by mounting an empty disk slice as a snapshot
of a currently mounted file system. The bitmap, blockmap and super-block are
initialized and then the currently mounted file system is frozen. After the file
system to be snapped is frozen, the snapshot is enabled and mounted and the
snapped file system is thawed. The snapshot appears as an exact image of the
snapped file system at the time the snapshot was made.
See “Freezing and thawing a file system” on page 65.
Initially, the snapshot file system satisfies read requests by finding the data on
the snapped file system and returning it to the requesting process. When an inode
update or a write changes the data in block n of the snapped file system, the old
data is first read and copied to the snapshot before the snapped file system is
updated. The bitmap entry for block n is changed from 0 to 1, indicating that the
data for block n can be found on the snapshot file system. The blockmap entry
for block n is changed from 0 to the block number on the snapshot file system
containing the old data.
A subsequent read request for block n on the snapshot file system will be satisfied
by checking the bitmap entry for block n and reading the data from the indicated
block on the snapshot file system, instead of from block n on the snapped file
system. This technique is called copy-on-write. Subsequent writes to block n on
the snapped file system do not result in additional copies to the snapshot file
system, since the old data only needs to be saved once.
All updates to the snapped file system for inodes, directories, data in files, extent
maps, and so forth, are handled in this fashion so that the snapshot can present
a consistent view of all file system structures on the snapped file system for the
time when the snapshot was created. As data blocks are changed on the snapped
file system, the snapshot gradually fills with data copied from the snapped file
system.
The amount of disk space required for the snapshot depends on the rate of change
of the snapped file system and the amount of time the snapshot is maintained. In
the worst case, the snapped file system is completely full and every file is removed
and rewritten. The snapshot file system would need enough blocks to hold a copy
of every block on the snapped file system, plus additional blocks for the data
structures that make up the snapshot file system. This is approximately 101
percent of the size of the snapped file system. Normally, most file systems do not
undergo changes at this extreme rate. During periods of low activity, the snapshot
should only require two to six percent of the blocks of the snapped file system.
During periods of high activity, the snapshot might require 15 percent of the
blocks of the snapped file system. These percentages tend to be lower for larger
file systems and higher for smaller ones.
Warning: If a snapshot file system runs out of space for changed data blocks, it is
disabled and all further attempts to access it fails. This does not affect the snapped
file system.
Chapter 7
Quotas
■ About quota limits
■ About quota files on &ProductNameLong;
■ About quota commands
■ About quota checking with Veritas File System
■ Using quotas
About quota limits

The &ProductNameLong; (VxFS) supports user and group and group and group
and group quotas. The quota system limits the use of two principal resources of
a file system: files and data blocks. For each of these resources, you can assign
quotas to individual users and groups and groups and groups and groups to limit
their usage.
You can set the following kinds of limits for each of the two resources:
hard limit An absolute limit that cannot be exceeded under any circumstances.
soft limit Must be lower than the hard limit, and can be exceeded, but only for
a limited time. The time limit can be configured on a per-file system
basis only. The VxFS default limit is seven days.
Soft limits are typically used when a user must run an application that could
generate large temporary files. In this case, you can allow the user to exceed the
quota limit for a limited time. No allocations are allowed after the expiration of
102 Quotas
About quota files on &ProductNameLong;
the time limit. Use the vxedquotavxedquotaedquotavxedquotavxedquota

command to set limits.
See ??? .
Although file and data block limits can be set individually for each user and
groupand groupand groupand group, the time limits apply to the file system as a
whole. The quota limit information is associated with user and group and group
and group and group IDs and is stored in a user or group or group or group or
group quota file.
See ??? .
The quota soft limit can be exceeded when VxFS preallocates space to a file.
See ??? .
About quota files on &ProductNameLong;

A quotas file (named quotas) must exist in the root directory of a file system for
any of the quota commands to work. For group quotas to work, there must be a
quotas.grp file.For group quotas to work, there must be a quotas.grp file.For group
quotas to work, there must be a quotas.grp file.For group quotas to work, there
must be a quotas.grp file.The files in the file system's mount point are referred
to as the externalquotas file. VxFS also maintains an internalquotas file for its
own use.
The quota administration commands read and write to the external quotas file to
obtain or change usage limits. VxFS uses the internal file to maintain counts of
data blocks and inodes used by each user. When quotas are turned on, the quota
limits are copied from the external quotas file into the internal quotas file. While
quotas are on, all the changes in the usage information and changes to quotas are
registered in the internal quotas file. When quotas are turned off, the contents
of the internal quotas file are copied into the external quotas file so that all data
between the two files is synchronized.
VxFS supports group quotas in addition to user quotas. Just as user quotas limit
file system resource (disk blocks and the number of inodes) usage on individual
users, group quotas specify and limit resource usage on a group basis. As with
user quotas, group quotas provide a soft and hard limit for file system resources.
If both user and group quotas are enabled, resource utilization is based on the
most restrictive of the two limits for a given user.
To distinguish between group and user quotas, VxFS quota commands use a –g
and –u option. The default is user quotas if neither option is specified. One
exception to this rule is when quotas are specified as a mount command option.
In this case, both user and group quotas are enabled. Support for group quotas
Quotas 103
About quota commands
also requires a separate group quotas file. The VxFS group quota file is named
quotas.grp. The VxFS user quotas file is named quotas. This name was used to
distinguish it from the quotas.user file used by other file systems under Solaris.
About quota commands

In general, quota administration for VxFS is performed using commands similar
to UFS quota commands. On Solaris, the available quota commands are
UFS-specific. That is, these commands work only on UFS file systems. For this
reason, VxFS supports a similar set of commands that work only for VxFS file
systems.
Note: Most of the quota commands in VxFS are similar to BSD quota commands.
However, the quotacheck command is an exception; VxFS does not support an
equivalent command.
See “About quota checking with &ProductNameLong;114.

VxFS supports the following quota-related commands:
vxedquota Edits quota limits for users and groups. The limit changes made by
vxedquota are reflected both in the internal quotas file and the external
quotas file.
vxrepquota Provides a summary of quotas and disk usage.
vxquot Provides file ownership and usage summaries.
vxquota Views quota limits and usage.
vxquotaon Turns quotas on for a mounted VxFS file system.
vxquotaoff Turns quotas off for a mounted VxFS file system.
Beside these commands, the VxFS mount command supports a special mount
option (–o quota), which can be used to turn on quotas at mount time.
For additional information on the quota commands, see the corresponding manual
pages.
Note:
Note:
104 Quotas
About quota checking with Veritas File System
Note:
Note: When VxFS file systems are exported via NFS, the VxFS quota commands
on the NFS client cannot query or edit quotas. You can use the VxFS quota
commands on the server to query or edit quotas.
About quota checking with Veritas File System

The standard practice with most quota implementations is to mount all file systems
and then run a quota check on each one. The quota check reads all the inodes on
disk and calculates the usage for each user and group. This can be time consuming,
and because the file system is mounted, the usage can change while quotacheck
is running.
VxFS does not support a quotacheck command. With VxFS, quota checking is
performed automatically, if necessary, at the time quotas are turned on. A quota
check is necessary if the file system has changed with respect to the usage
information as recorded in the internal quotas file. This happens only if the file
system was written with quotas turned off, or if there was structural damage to
the file system that required a full file system check.
See the fsck_vxfs(1M) manual page.
A quota check generally reads information for each inode on disk and rebuilds
the internal quotas file. It is possible that while quotas were not on, quota limits
were changed by the system administrator. These changes are stored in the
external quotas file. As part of enabling quotas processing, quota limits are read
from the external quotas file into the internal quotas file.
Using quotas
The VxFS quota commands are used to manipulate quotas.
Turning on quotas
To use the quota functionality on a file system, quotas must be turned on. You
can turn quotas on at mount time or after a file system is mounted.
Quotas 105
Using quotas
Note: Before turning on quotas, the root directory of the file system must contain
a file for user quotas named quotas and a file for group quotas named quotas.grp
and a file for group quotas named quotas.grp and a file for group quotas named
quotas.grp and a file for group quotas named quotas.grp owned by root.
To turn on quotas
1 To turn on user and groupuser and groupuser and groupuser and group quotas
for a VxFS file system, enter:
# vxquotaonvxquotaonquotaonvxquotaonvxquotaon /mount_point
2 To turn on only user quotas for a VxFS file system, enter:
# vxquotaon –u /mount_point
3 To turn on only group quotas for a VxFS file system, enter:
# vxquotaon –g /mount_point
4
5
6
7
8
9
Turning on quotas at mount time

Quotas can be turned on with the mount command when you mount a file system.
To turn on quotas at mount time
1 To turn on user or group user or group user or group user or group quotas
for a file system at mount time, enter:
# mount –FFFtV vxfs –o quota special /mount_point
2 To turn on only user quotas, enter:
# mount –F vxfs –o usrquota special /mount_point

106 Quotas
Using quotas
3 To turn on only group quotas, enter:
# mount –F vxfs –o grpquota special /mount_point
4
5
6
7
8
9
Editing user and group and group and group and group quotas
You can set up user and group quotas using the vxedquota command. You must
have superuser privileges to edit quotas.
vxedquota creates a temporary file for the given user; this file contains on-disk
quotas for each mounted file system that has a quotas file. It is not necessary that
quotas be turned on for vxedquota to work. However, the quota limits are
applicable only after quotas are turned on for a given file system.
To edit quotas
1 Specify the –u option to edit the quotas of one or more users specified by
username:
# vxedquota [–u] username
Editing the quotas of one or more users is the default behavior if the –u option
is not specified.
2 Specify the –g option to edit the quotas of one or more groups specified by
groupname:
# vxedquota –g groupname
1
2
◆
1
2
Quotas 107
Using quotas
1
2
Modifying time limits

The soft and hard limits can be modified or assigned values. For any user or group,
usage can never exceed the hard limit after quotas are turned on.
Modified time limits apply to the entire file system and cannot be set selectively
for each user or group.
To modify time limits
1 Specify the -t option to modify time limits for any user:
# vxedquota [–u] –t
2 Specify the -g and -t options to modify time limits for any group:
# vxedquota –g –t
Viewing disk quotas and usage

Use the vxquota command to view a user's or group's disk quotas and usage on
VxFS file systems.
To display disk quotas and usage
1 To display a user's quotas and disk usage on all mounted VxFS file systems
where the quotas file exists, enter:
# vxquota –v [–u] username
2 To display a group's quotas and disk usage on all mounted VxFS file systems
where the quotas.grp file exists, enter:
# vxquota –v –g groupname
1
2
3
1
2
108 Quotas
Using quotas
1
2
Displaying blocks owned by users or groups

Use the vxquot command to display the number of blocks owned by each user or
group in a file system.
To display the number of blocks owned by users or groups
1 To display the number of files and the space owned by each user, enter:
# vxquot [–u] –f filesystem
2 To display the number of files and the space owned by each group, enter:
# vxquot –g –f filesystem
1
2
◆
1
2
1
2
Turning off quotas

Use the vxquotaoff command to turn off quotas.
To turn off quotas
1 To turn off quotas for a mounted file system, enter:
# vxquotaoff /mount_point
2 To turn off only user quotas for a VxFS file system, enter:
# vxquotaoff –u /mount_point
3 To turn off only group quotas for a VxFS file system, enter:
# vxquotaoff –g /mount_point
Chapter 8
File Change Log
■ About the File Change Log file
■ File Change Log administrative interface
■ File Change Log programmatic interface
■ Reverse path name lookup
About the File Change Log file

File Change Log records file system changes such as creates, links, unlinks,
renaming, data appended, data overwritten, data truncated, extended attribute
modifications, holes punched, and miscellaneous file property updates.
Note: FCL is supported only on disk layout Version 6 and later.
FCL stores changes in a sparse file in the file system namespace. The FCL log file
is located in mount_point/lost+found/changelog. The FCL file behaves like a
regular file, but some operations are prohibited. The standard system calls open(2),
lseek(2), read(2) and close(2) can access the data in the FCL, while the write(2),
mmap(2) and rename(2) calls are not allowed.
Warning: In future VxFS releases, the FCL file might be pulled out of the namespace
such that these standard system calls will no longer work. Therefore, it is
recommended that all new applications be developed using the programmatic
interface.
110 File Change Log
File Change Log administrative interface
The FCL log file contains both the information about the FCL, which is stored in
the FCL superblock, and the changes to files and directories in the file system,
which is stored as FCL records.
See “File Change Log programmatic interface” on page 112.
In 4.1, the structure of the File Change Log file was exposed through the
/opt/VRTS/include/sys/fs/fcl.h header file. In this release, the internal
structure of the FCL file is opaque. The recommended mechanism to access the
FCL is through the API described by the /opt/VRTSfssdk/5.0/include/vxfsutil.h
header file.
The /opt/VRTS/include/sys/fs/fcl.h header file is included in this release to
ensure that applications accessing the FCL with the 4.1 header file do not break.
New applications should use the new FCL API described in
/opt/VRTSfssdk/5.0/include/vxfsutil.h. Existing applications should also be
modified to use the new FCL API.
With the addition of new record types, the FCL version in this release has been
updated to 4. To provide backward compatibility for the existing applications,
this release supports multiple FCL versions. Users now have the flexibility of
specifying of specifying the FCL version for new FCLs. The default FCL version is
4.
See the fcladm(1M) man page.

The FCL can be set up and tuned through the fcladm and vxtunefs VxFS
administrative commands.
See the fcladm(1M) and vxtunefs(1M) manual pages.
The FCL set parameters are for fcladm:
set Enables the recording of the audit, open, close, and stats events
in the File Change Log file. Setting the audit option enables all
events to be recorded in the FCL file when the command is issued.
Setting the audit option also lists the struct fcl_accessinfo
identifier, which shows the user ID. The open option enables all
files opened when the command is issued along with the command
names to be recorded in the FCL file. The close option allows the
recording of all files that are closed in the FCL file. The stats option
enables the statistics of all events to be recorded to the FCL.
clear Disables the recording of the audit, open, close, and stats events
after it has been set.
File Change Log 111
The FCL tunable parameters for vxtunefs are:
fcl_keeptime Specifies the duration in seconds that FCL records stay in the FCL
file before they can be purged. The first records to be purged are
the oldest ones, which are located at the beginning of the file.
Additionally, records at the beginning of the file can be purged if
allocation to the FCL file exceeds fcl_maxalloc bytes. The
default value is 0. Note that fcl_keeptime takes precedence
over fcl_maxalloc. No hole is punched if the FCL file exceeds
fcl_maxalloc bytes but the life of the oldest record has not
reached fcl_keeptime seconds.
fcl_maxalloc Specifies the maximum number of spaces in bytes to be allocated

to the FCL file. When the space allocated exceeds fcl_maxalloc,
a hole is punched at the beginning of the file. As a result, records
are purged and the first valid offset (fc_foff) is updated. In addition,
fcl_maxalloc may be violated if the oldest record has not
reached fcl_keeptime.
The minimum value of fcl_maxalloc is 4 MB. The default value

is fs_size/33.
fcl_winterval Specifies the time in seconds that must elapse before the FCL
records an overwrite, extending write, or a truncate. This helps
to reduce the number of repetitive records in the FCL. The
fcl_winterval timeout is per inode. If an inode happens to go
out of cache and returns, its write interval is reset. As a result,
there could be more than one write record for that file in the same
write interval. The default value is 3600 seconds.
fcl_ointerval The time interval in seconds within which subsequent opens of a

file do not produce an additional FCL record. This helps to reduce
the number of repetitive records logged in the FCL file. If the
tracking of access information is also enabled, a subsequent file
open even within the fcl_ointerval may produce a record, if it is
opened by a different user. Similarly, if the inode is bumped out
of cache, this may also produce more than one record within the
same open interval.
The default value is 600 sec.
Either or both fcl_maxalloc and fcl_keeptime must be set to activate the FCL
feature. The following are examples of using the FCL administration command.
To activate FCL for a mounted file system, enter:
# fcladm on mount_point
To deactivate the FCL for a mounted file system, enter:

112 File Change Log
File Change Log programmatic interface
# fcladm off mount_point
To remove the FCL file for a mounted file system, on which FCL must be turned
off, enter:
# fcladm rm mount_point
To obtain the current FCL state for a mounted file system, enter:
# fcladm state mount_point
Print the on-disk FCL super-block in text format to obtain information about the
FCL file by using offset 0. Because the FCL on-disk super-block occupies the first
block of the FCL file, the first and last valid offsets into the FCL file can be
determined by reading the FCL super-block and checking the fc_foff field. Enter:
# fcladm print 0 mount_point
To print the contents of the FCL in text format, of which the offset used must be
32-byte aligned, enter:
# fcladm print offset mount_point

VxFS provides an enhanced API to simplify reading and parsing the FCL file in
two ways:
Simplified reading The API simplifies user tasks by reducing additional code needed
to parse FCL file entries. In 4.1, to obtain event information such
as a remove or link, the user was required to write additional code
to get the name of the removed or linked file. In this release, the
API allows the user to directly read an assembled record. The API
also allows the user to specify a filter to indicate a subset of the
event records of interest.
Backward Providing API access for the FCL feature allows backward
compatibility compatibility for applications. The API allows applications to
parse the FCL file independent of the FCL layout changes. Even if
the hidden disk layout of the FCL changes, the API automatically
translates the returned data to match the expected output record.
As a result, the user does not need to modify or re-compile the
application due to changes in the on-disk FCL layout.
See “Reverse path name lookup” on page 114.

File Change Log 113
The following sample code fragment reads the FCL superblock, checks that the
state of the FCL is VX_FCLS_ON, issues a call to vxfs_fcl_sync to obtain a finishing
offset to read to, determines the first valid offset in the FCL file, then reads the
entries in 8K chunks from this offset. The section process fcl entries is what an
application developer must supply to process the entries in the FCL.
#include <stdint.h>
#include <stdio.h>
#include <stdlib.h>
#include <sys/types.h>
#include <sys/fcntl.h>
#include <errno.h>
#include <fcl.h>
#include <vxfsutil.h>
#define FCL_READSZ 8192
char* fclname = "/mnt/lost+found/changelog";
int read_fcl(fclname) char* fclname;
{
struct fcl_sb fclsb;
uint64_t off, lastoff;
size_t size;
char buf[FCL_READSZ], *bufp = buf;
int fd;
int err = 0;
if ((fd = open(fclname, O_RDONLY)) < 0) {
return ENOENT;
}
if ((off = lseek(fd, 0, SEEK_SET)) != 0) {
close(fd);
return EIO;
}
size = read(fd, &fclsb, sizeof (struct fcl_sb));
if (size < 0) {
close(fd);
return EIO;
}
if (fclsb.fc_state == VX_FCLS_OFF) {
close(fd);
return 0;
}
if (err = vxfs_fcl_sync(fclname, &lastoff)) {
close(fd);
return err;
114 File Change Log
Reverse path name lookup
}
if ((off = lseek(fd, off_t, uint64_t)) != uint64_t) {
close(fd);
return EIO;
}
while (off < lastoff) {
if ((size = read(fd, bufp, FCL_READSZ)) <= 0) {
close(fd);
return errno;
}
/* process fcl entries */
off += size;
}
close(fd);
return 0;
}

The reverse path name lookup feature obtains the full path name of a file or
directory from the inode number of that file or directory. The inode number is
provided as an argument to the vxlsino administrative command, or the
vxfs_inotopath_gen(3) application programming interface library function.
The reverse path name lookup feature can be useful for a variety of applications,
such as for clients of the VxFS File Change Log feature, in backup and restore
utilities, and for replication products. Typically, these applications store
information by inode numbers because a path name for a file or directory can be
very long, thus the need for an easy method of obtaining a path name.
An inode is a unique identification number for each file in a file system. An inode
contains the data and metadata associated with that file, but does not include the
file name to which the inode corresponds. It is therefore relatively difficult to
determine the name of a file from an inode number. The ncheck command provides
a mechanism for obtaining a file name from an inode identifier by scanning each
directory in the file system, but this process can take a long period of time. The
VxFS reverse path name lookup feature obtains path names relatively quickly.
Note: Because symbolic links do not constitute a path to the file, the reverse path
name lookup feature cannot track symbolic links to files.
File Change Log 115
Because of the possibility of errors with processes renaming or unlinking and

creating new files, it is advisable to perform a lookup (or open) with the path name
and verify that the inode number matches the path names obtained.
See the vxlsino(1M), vxfs_inotopath_gen(3), and vxfs_inotopath(3) manual pages.
116 File Change Log
Chapter 9
Multi-volume file systems
■ About multi-volume support
■ About volume types
■ Features implemented using multi-volume support
■ About volume sets
■ Creating multi-volume file systems
■ Converting a single volume file system to a multi-volume file system
■ Removing a volume from a multi-volume file system
■ About allocation policies
■ Assigning allocation policies
■ Querying allocation policies
■ Assigning pattern tables to directories
■ Assigning pattern tables to file systems
■ Allocating data
■ Volume encapsulation
■ Reporting file extents
■ Load balancing
■ Converting a multi-volume file system to a single volume file system

118 Multi-volume file systems
About multi-volume support
About multi-volume support

VxFS provides support for multi-volume file systems when used in conjunction
with the Veritas Volume Manager. Using multi-volume support (MVS), a single
file system can be created over multiple volumes, each volume having its own
properties. For example, it is possible to place metadata on mirrored storage while
placing file data on better-performing volume types such as RAID-1+0 (striped
and mirrored).
The MVS feature also allows file systems to reside on different classes of devices,
so that a file system can be supported from both inexpensive disks and from
expensive arrays. Using the MVS administrative interface, you can control which
data goes on which volume types.
Note: MVS is available only on file systems using disk layout Version 6 or later.
See “About disk layouts” on page 277.
About volume types

VxFS utilizes two types of volumes, one of which contains only data, referred to
as dataonly, and the other of which can contain metadata or data, referred to as
metadataok.
Data refers to direct extents, which contain user data, of regular files and named
data streams in a file system.
Metadata refers to all data allocated when the file system was created with the
mkfs command or allocated at a later time. Metadata includes extents allocated
to structural files, blocks reserved for the super block, the volume label, indirect
extents, extents belonging to the File Change Log file, the history log file, extended
attributes, directories, access control lists, and so on.
A volume availability flag is set to specify if a volume is dataonly or metadataok.
The volume availability flag can be set, cleared, and listed with the fsvoladm
command.
See the fsvoladm(1M) manual page.
Features implemented using multi-volume support

The following features can be implemented using multi-volume support:
■ Controlling where files are stored can be selected at multiple levels so that
specific files or file hierarchies can be assigned to different volumes. This
Multi-volume file systems 119
Features implemented using multi-volume support
functionality is available in the Veritas File System Dynamic Storage Tiering

(DST) feature.
■ Placing the VxFS intent log on its own volume to minimize disk head movement
and thereby increase performance. This functionality can be used to migrate
from the Veritas QuickLog™ feature.
■ Separating Storage Checkpoints so that data allocated to a Storage Checkpoint
is isolated from the rest of the file system.
■ Separating metadata from file data.
■ Encapsulating volumes so that a volume appears in the file system as a file.
This is particularly useful for databases that are running on raw volumes.
■ Guaranteeing the availability of some volumes even when others are
unavailable.
To use the multi-volume file system features, Veritas Volume Manager must be
installed and the volume set feature must be accessible.
Volume availability
MVS guarantees the availability of some volumes even when others are unavailable.
This allows you to mount a multi-volume file system even if one or more
component dataonly volumes are missing.
The volumes are separated by whether metadata is allowed on the volume. An
I/O error on a dataonly volume does not affect access to any other volumes. All
VxFS operations that do not access the missing dataonly volume function normally,
including:
■ Mounting the multi-volume file system, regardless if the file system is read-only
or read/write.
■ Kernel operations.
■ Performing a fsck replay. Logged writes are converted to normal writes if the
corresponding volume is dataonly.
■ Performing a full fsck.
■ Using all other commands that do not access data on a missing volume.
Some operations that could fail if a dataonly volume is missing include:
■ Reading or writing file data if the file's data extents were allocated from the
missing dataonly volume.
■ Using the vxdump command.
About volume sets
Volume availability is supported only on a file system with disk layout Version 7
or later.
Note: Do not mount a multi-volume system with the ioerror=disable or

ioerror=wdisable mount options if the volumes have different availability
properties. Symantec recommends the ioerror=mdisable mount option for cluster
mounts and ioerror=mwdisable for local mounts.
About volume sets

Veritas Volume Manager exports a data object called a volume set. Unlike the
traditional Volume Manager volume, which can be used for raw I/O access or to
contain a file system, a volume set is a container for multiple different volumes.
Each volume can have its own geometry.
The Volume Manager vxvset command is used to create and manage volume sets.
Volume sets cannot be empty. When the last entry is removed, the volume set
itself is removed.
Creating and managing volume sets

The following command examples show how to create and manage volume sets.
To create and manage volume sets
1 Create a new volume set from vol1:
# vxassist make vol1 10m

# vxvset make myvset vol1
2 Create two new volumes and add them to the volume set:

# vxvset addvol myvset vol2
Creating multi-volume file systems
3 List the component volumes of the previously created volume set:
# vxvset list myvset

VOLUME INDEX LENGTH STATE CONTEXT
vol1 0 20480 ACTIVE -
4 Use the ls command to see that when a volume set is created, the volumes
contained by the volume set are removed from the namespace and are instead
accessed through the volume set name:
# ls -l /dev/vx/rdsk/rootdg/myvset
1 root root 108,70009 May 21 15:37 /dev/vx/rdsk/rootdg/myvset
5 Create a volume, add it to the volume set, and use the ls command to see that
when a volume is added to the volume set, it is no longer visible in the
namespace:

# ls -l /dev/vx/rdsk/rootdg/vol4
crw-- 1 root root 108,70012 May 21 15:43
/dev/vx/rdsk/rootdg/vol4
# ls -l /dev/vx/rdsk/rootdg/vol4
/dev/vx/rdsk/rootdg/vol4: No such file or directory

When a multi-volume file system is created, all volumes are dataonly, except
volume zero. The volume availability flag of volume zero cannot be set to dataonly.
As metadata cannot be allocated from dataonly volumes, enough metadata space
should be allocated using metadataok volumes. The “file system out of space”
error occurs if there is insufficient metadata space available, even if the df
command shows that there is free space in the file system. The fsvoladm command
can be used to see the free space in each volume and set the availability flag of
the volume.
With the exception of adding and deleting volumes, the file system commands
operate the same on volumes within a volume set.
Example of creating a multi-volume file system

The following procedure is an example of creating a multi-volume file system.
To create a multi-volume file system
1 After a volume set is created, create a VxFS file system by specifying the
volume set name as an argument to mkfs:
# mkfs -F vxfs /dev/vx/rdsk/rootdg/myvset

version 7 layout
327680 sectors, 163840 blocks of size 1024,
log size 1024 blocks largefiles supported
After the file system is created, VxFS allocates space from the different
volumes within the volume set.
2 List the component volumes of the volume set using of the fsvoladm command:
# mount -F vxfs /dev/vx/dsk/rootdg/myvset /mnt1

0 10240 1280 8960 vol1
1 51200 16 51184 vol2
2 51200 16 51184 vol3
3 51200 16 51184 vol4
3 Add a new volume by adding the volume to the volume set, then adding the
volume to the file system:

# fsvoladm add /mnt1 vol5 50m
0 10240 1300 8940 vol1
1 51200 16 51184 vol2
2 51200 16 51184 vol3
3 51200 16 51184 vol4
4 51200 16 51184 vol5
Converting a single volume file system to a multi-volume file system
4 List the volume availability flags using the fsvoladm command:
# fsvoladm queryflags /mnt1

volname flags
vol1 metadataok
vol2 dataonly
vol3 dataonly
vol4 dataonly
vol5 dataonly
5 Increase the metadata space in the file system using the fsvoladm command:
# fsvoladm clearflags dataonly /mnt1 vol2

# fsvoladm queryflags /mnt1
volname flags
vol1 metadataok
vol2 metadataok
vol3 dataonly
vol4 dataonly
vol5 dataonly
Converting a single volume file system to a

multi-volume file system
The following procedure converts a traditional, single volume file system, /mnt1,
on a single volume vol1 in the diskgroup dg1 to a multi-volume file system.
To convert a single volume file system
1 Determine the version of the volume's diskgroup:
# vxdg list dg1 | grep version: | awk '{ print $2 }'

105
2 If the version is less than 110, upgrade the diskgroup:
# vxdg upgrade dg1
3 Determine the disk layout version of the file system:
# vxupgrade /mnt1
Version 5
Removing a volume from a multi-volume file system
4 If the disk layout is version is less than 6, upgrade to Version 7:
# vxupgrade -n 7 /mnt1
5 Unmount the file system:
# umount /mnt1
6 Convert the volume into a volume set:
# vxvset -g dg1 make vset1 vol1
7 Edit the /etc/vfstab file to replace the volume device name, vol1, with the
volume set name, vset1.
8 Mount the file system:
# mount -F vxfs /dev/vx/dsk/dg1/vset1 /mnt1
9 As necessary, create and add volumes to the volume set:
# vxassist -g dg1 make vol2 256M

# vxvset -g dg1 addvol vset1 vol2
10 As necessary, set the placement class tags on the volumes:

# vxvoladm -g dg1 settag vol1 vxfs.placement_class.tier1
# vxvoladm -g dg1 settag vol2 vxfs.placement_class.tier2
11 Add the volumes to the file system:

Removing a volume from a multi-volume file system

Use the fsvoladm remove command to remove a volume from a multi-volume file
system. The fsvoladm remove command fails if an allocation policy exists that
has only a target device.
Forcibly removing a volume

If you must forcibly remove a volume from a file system, such as if a volume is
permanently destroyed and you want to clean up the dangling pointers to the lost
About allocation policies
volume, use the fsck -o zapvol=volname command. The zapvol option performs
a full file system check and zaps all inodes that refer to the specified volume. The
fsck command prints the inode numbers of all files that the command destroys;
the file names are not printed. The zapvol option only affects regular files if used
on a dataonly volume. However, it could destroy structural files if used on a
metadataok volume, which can make the file system unrecoverable. Therefore,
the zapvol option should be used with caution on metadataok volumes.
About allocation policies

To make full use of multi-volume support features, VxFS provides support for
allocation policies that allow files or groups of files to be assigned to specified
volumes within the volume set.
A policy specifies a list of volumes and the order in which to attempt allocations.
A policy can be assigned to a file, a file system, or a Storage Checkpoint created
from a file system. When policies are assigned to objects in the file system, you
must specify how the policy maps to both metadata and file data. For example, if
a policy is assigned to a single file, the file system must know where to place both
the file data and metadata. If no policies are specified, the file system places data
randomly.
Assigning allocation policies

The following example shows how to assign allocation policies. The example
volume set contains two volumes from different classes of storage.
Querying allocation policies
To assign allocation policies

1 List the volumes in the volume set:
# vxvset -g rootdg list myvset

2 Create a file system on the myvset volume set and mount it:

version 7 layout
log size 1024 blocks
largefiles supported
3 Define two allocation policies called datapolicy and metadatapolicy to refer

to the vol1 and vol2 volumes:
# fsapadm define /mnt1 datapolicy vol1

# fsapadm define /mnt1 metadatapolicy vol2
4 Assign the policies at the file system level. The data policy must be specified
before the metadata policy:
# fsapadm assignfs /mnt1 datapolicy metadatapolicy

0 51200 1250 49950 vol1
1 51200 16 51184 vol2
The assignment of the policies on a file system-wide basis ensures that any
metadata allocated is stored on the device with the policy metadatapolicy
(vol2) and all user data is be stored on vol1 with the associated datapolicy
policy.
Querying allocation policies

Querying an allocation policy displays the definition of the allocation policy.
The following example shows how to query allocation policies. The example volume
set contains two volumes from different classes of storage.
Assigning pattern tables to directories
To query allocation policies

◆ Query the allocation policy:
# fsapadm query /mnt1 datapolicy
Assigning pattern tables to directories

A pattern table contains patterns against which a file's name and creating process'
UID and GID are matched as a file is created in a specified directory. The first
successful match is used to set the allocation policies of the file, taking precedence
over inheriting per-file allocation policies.
See the fsapadm(1M) manual page.
The following example shows how to assign pattern tables to a directory in a
volume set that contains two volumes from different classes of storage. The
pattern table matches all files created in the directory dir1 with the .mp3 extension
for any user or group ID and assigns the mp3data data policy and mp3meta metadata
policy.
To assign pattern tables to directories
1 Define two allocation policies called mp3data and mp3meta to refer to the vol1
and vol2 volumes:
# fsapadm define /mnt1 mp3data vol1

# fsapadm define /mnt1 mp3meta vol2
2 Assign the pattern table:
# fsapadm assignfilepat dir1 *.mp3///mp3data/mp3meta/
Assigning pattern tables to file systems

A pattern table contains patterns against which a file's name and creating process'
UID and GID are matched as a file is created in a directory. If the directory does
not have its own pattern table or an inheritable allocation policy, the file system's
pattern table takes effect. The first successful match is used to set the allocation
policies of the file.
See the fsapadm(1M) manual page.
Allocating data
The following example shows how to assign pattern tables to a file system in a
volume set that contains two volumes from different classes of storage. The
pattern table is contained within the pattern file mypatternfile.
To assign pattern tables to directories
1 Define two allocation policies called mydata and mymeta to refer to the vol1
and vol2 volumes:
# fsapadm define /mnt1 mydata vol1

# fsapadm define /mnt1 mymeta vol2
2 Assign the pattern table:
# fsapadm assignfspat -F mypatternfile /mnt1
Allocating data
The following script creates a large number of files to demonstrate the benefit of
allocating data:
i=1
while [ $i -lt 1000 ]
do
dd if=/dev/zero of=/mnt1/$i bs=65536 count=1
i=‘expr $i + 1‘
done
Before the script completes, vol1 runs out of space even though space is still
available on the vol2 volume:

0 51200 51200 0 vol1
1 51200 221 50979 vol2
The solution is to assign an allocation policy that allocates user data from the
vol1 volume to vol2 if space runs out.
You must have system administrator privileges to create, remove, change policies,
or set file system or Storage Checkpoint level policies. Users can assign a
pre-existing policy to their files if the policy allows that.
Policies can be inherited for new files. A file will inherit the allocation policy of
the directory in which it resides if you run the fsapadm assignfile -f inherit
command on the directory.
Volume encapsulation
Allocating data from vol1 to vol2

◆ Assign an allocation policy that allocates user data from vol1 to vol2 if space
runs out on vol1:
# fsapadm define /mnt1 datapolicy vol1 vol2
Multi-volume support enables the ability to encapsulate an existing raw volume
and make the volume contents appear as a file in the file system.
Encapsulating a volume involves the following actions:
■ Adding the volume to an existing volume set.
■ Adding the volume to the file system using fsvoladm
Encapsulating a volume
The following example illustrates how to encapsulate a volume.
To encapsulate a volume
1 List the volumes:

The volume set has two volumes.

2 Create a third volume and copy the passwd file to the third volume:
# vxassist make dbvol 100m

# dd if=/etc/passwd of=/dev/vx/rdsk/rootdg/dbvol count=1
1+0 records in
1+0 records out
The third volume will be used to demonstrate how the volume can be accessed
as a file, as shown later.
3 Create a file system on the volume set:

version 7 layout
log size 1024 blocks
4 Mount the volume set:
5 Add the new volume to the volume set:
# vxvset addvol myvset dbvol
6 Encapsulate dbvol:
# fsvoladm encapsulate /mnt1/dbfile dbvol 100m

# ls -l /mnt1/dbfile
-rw------- 1 root other 104857600 May 22 11:30 /mnt1/dbfile
7 Examine the contents of dbfile to see that it can be accessed as a file:
# head -2 /mnt1/dbfile
root:x:0:1:Super-User:/:/sbin/sh
daemon:x:1:1::/:
The passwd file that was written to the raw volume is now visible in the new
file.
Note: If the encapsulated file is changed in any way, such as if the file is
extended, truncated, or moved with an allocation policy or resized volume,
or the volume is encapsulated with a bias, the file cannot be de-encapsulated.
Deencapsulating a volume
The following example illustrates how to deencapsulate a volume.
Reporting file extents
To deencapsulate a volume
1 List the volumes:

dbvol 2 102400 ACTIVE -
The volume set has three volumes.

2 Deencapsulate dbvol:
# fsvoladm deencapsulate /mnt1/dbfile
Reporting file extents

MVS feature provides the capability for file-to-volume mapping and volume-to-file
mapping via the fsmap and fsvmap commands. The fsmap command reports the
volume name, logical offset, and size of data extents, or the volume name and size
of indirect extents associated with a file on a multi-volume file system. The fsvmap
command maps volumes to the files that have extents on those volumes.
See the fsmap(1M) and fsvmap(1M) manual pages.
The fsmap command requires open() permission for each file or directory specified.
Root permission is required to report the list of files with extents on a particular
volume.
Examples of reporting file extents

The following examples show typical uses of the fsmap and fsvmap commands.
Using the fsmap command
◆ Use the find command to descend directories recursively and run fsmap on
the list of files:
# find . | fsmap -
Volume Extent Type File
vol2 Data ./file1
vol1 Data ./file2
Load balancing
Using the fsvmap command

1 Report the extents of files on multiple volumes:
# fsvmap /dev/vx/rdsk/fstest/testvset vol1 vol2

vol1 /.
vol1 /ns2
vol1 /ns3
vol1 /file1
vol2 /file1
vol2 /file2
2 Report the extents of files that have either data or metadata on a single volume
in all Storage Checkpoints, and indicate if the volume has file system metadata:
# fsvmap -mvC /dev/vx/rdsk/fstest/testvset vol1

Meta Structural vol1 //volume has filesystem metadata//
Data UNNAMED vol1 /.
Data UNNAMED vol1 /ns2
Data UNNAMED vol1 /ns3
Data UNNAMED vol1 /file1
Meta UNNAMED vol1 /file1
Load balancing
An allocation policy with the balance allocation order can be defined and assigned
to files that must have their allocations distributed at random between a set of
specified volumes. Each extent associated with these files are limited to a maximum
size that is defined as the required chunk size in the allocation policy. The
distribution of the extents is mostly equal if none of the volumes are full or
disabled.
Load balancing allocation policies can be assigned to individual files or for all files
in the file system. Although intended for balancing data extents across volumes,
a load balancing policy can be assigned as a metadata policy if desired, without
any restrictions.
Load balancing
Note: If a file has both a fixed extent size set and an allocation policy for load
balancing, certain behavior can be expected. If the chunk size in the allocation
policy is greater than the fixed extent size, all extents for the file are limited by
the chunk size. For example, if the chunk size is 16 MB and the fixed extent size
is 3 MB, then the largest extent that satisfies both the conditions is 15 MB. If the
fixed extent size is larger than the chunk size, all extents are limited to the fixed
extent size. For example, if the chunk size is 2 MB and the fixed extent size is 3
MB, then all extents for the file are limited to 3 MB.
Defining and assigning a load balancing allocation policy

The following example defines a load balancing policy and assigns the policy to
the file, /mnt/file.db.
To define and assign the policy
1 Define the policy by specifying the -o balance and -c options:
# fsapadm define -o balance -c 2m /mnt loadbal vol1 vol2 vol3 vol4
2 Assign the policy:
# fsapadm assign /mnt/filedb loadbal meta
Rebalancing extents
Extents can be rebalanced by strictly enforcing the allocation policy. Rebalancing
is generally required when volumes are added or removed from the policy or when
the chunk size is modified. When volumes are removed from the volume set, any
extents on the volumes being removed are automatically relocated to other volumes
within the policy.
The following example redefines a policy that has four volumes by adding two
new volumes, removing an existing volume, and enforcing the policy for
rebalancing.
Converting a multi-volume file system to a single volume file system
To rebalance extents
1 Define the policy by specifying the -o balance and -c options:
# fsapadm define -o balance -c 2m /mnt loadbal vol1 vol2 vol4 \

vol5 vol6
2 Assign the policy:
# fsapadm enforcefile -f strict /mnt/filedb
Converting a multi-volume file system to a single

volume file system
Because data can be relocated among volumes in a multi-volume file system, you
can convert a multi-volume file system to a traditional, single volume file system
by moving all file system data onto a single volume. Such a conversion is useful
to users who would like to try using a multi-volume file system or Dynamic Storage
Tiering, but are not committed to using a multi-volume file system permanently.
There are three restrictions to this operation:
■ The single volume must be the first volume in the volume set
■ The first volume must have sufficient space to hold all of the data and file
system metadata
■ The volume cannot have any allocation policies that restrict the movement of
data
Converting to a single volume file system

The following procedure converts an existing multi-volume file system, /mnt1,
of the volume set vset1, to a single volume file system, /mnt1, on volume vol1 in
diskgroup dg1.
Note: Steps 5, 6, and 8 are optional, and can be performed if you prefer to remove
the wrapper of the volume set object.
Converting to a single volume file system

1 Determine if the first volume in the volume set, which is identified as device
number 0, has the capacity to receive the data from the other volumes that
will be removed:
# df /mnt1
/mnt1 (/dev/vx/dsk/dg1/vol1):16777216 blocks 3443528 files
2 If the first volume does not have sufficient capacity, grow the volume to a
sufficient size:
# fsvoladm resize /mnt1 vol1 150g
3 Remove all existing allocation policies:
# fsppadm unassign /mnt1
4 Remove all volumes except the first volume in the volume set:
# fsvoladm remove /mnt1 vol2

# vxvset -g dg1 rmvol vset1 vol2
# fsvoladm remove /mnt1 vol3
# vxvset -g dg1 rmvol vset1 vol3
Before removing a volume, the file system attempts to relocate the files on
that volume. Successful relocation requires space on another volume, and no
allocation policies can be enforced that pin files to that volume. The time for
the command to complete is proportional to the amount of data that must be
relocated.
5 Unmount the file system:
# umount /mnt1
6 Remove the volume from the volume set:
# vxvset -g dg1
7 Edit the /etc/vfstab file to replace the volume set name, vset1, with the
volume device name, vol1.
8 Mount the file system:
# mount -F vxfs /dev/vx/dsk/dg1/vol1 /mnt1

Chapter 10
Dynamic Storage Tiering
■ About Dynamic Storage Tiering
■ Placement classes
■ Administering placement policies
■ File placement policy grammar
■ File placement policy rules
■ Calculating I/O temperature and access temperature
■ Multiple criteria in file placement policy rule statements
■ File placement policy rule and statement ordering
■ File placement policies and extending files
About Dynamic Storage Tiering

VxFS uses multi-tier online storage via the Dynamic Storage Tiering (DST) feature,
which functions on top of multi-volume file systems. Multi-volume file systems
are file systems that occupy two or more virtual volumes. The collection of volumes
is known as a volume set, and is made up of disks or disk array LUNs belonging
to a single Veritas Volume Manager (VxVM) disk group. A multi-volume file system
presents a single name space, making the existence of multiple volumes
transparent to users and applications. Each volume retains a separate identity
for administrative purposes, making it possible to control the locations to which
individual files are directed.
See “About multi-volume support” on page 118.
138 Dynamic Storage Tiering
About Dynamic Storage Tiering
Note: Some of the commands have changed or removed between the 4.1 release
and the 5.0 release to make placement policy management more user-friendly.
The following are the commands that have been removed: fsrpadm, fsmove, and
fssweep. The output of the queryfile, queryfs, and list options of the fsapadm
command now print the allocation order by name instead of number.
DST allows administrators of multi-volume VxFS file systems to manage the

placement of files on individual volumes in a volume set by defining placement
policies that control both initial file location and the circumstances under which
existing files are relocated. These placement policies cause the files to which they
apply to be created and extended on specific subsets of a file system's volume set,
known as placement classes. The files are relocated to volumes in other placement
classes when they meet the specified naming, timing, access rate, and storage
capacity-related conditions.
You make a VxVM volume part of a placement class by associating a volume tag
with it. For file placement purposes, VxFS treats all of the volumes in a placement
class as equivalent, and balances space allocation across them. A volume may
have more than one tag associated with it. If a volume has multiple tags, the
volume belongs to multiple placement classes and is subject to allocation and
relocation policies that relate to any of the placement classes. Multiple tagging
should be used carefully.
See “Placement classes” on page 139.
VxFS imposes no capacity, performance, availability, or other constraints on
placement classes. Any volume may be added to any placement class, no matter
what type the volume has nor what types other volumes in the class have. However,
a good practice is to place volumes of similar I/O performance and availability in
the same placement class.
Note: Dynamic Storage Tiering is a licensed feature. You must purchase a separate
license key for DST to operate. See the Veritas Storage Foundation Release Notes.
The Using Dynamic Storage Tiering Symantec Yellow Book provides additional
information regarding the Dynamic Storage Tiering feature, including the value
of DST and best practices for using DST. You can download Using Dynamic Storage
Tiering from the following webpage:
http://www.symantec.com/enterprise/yellowbooks/index.jsp
Dynamic Storage Tiering 139
Placement classes
Placement classes
A placement class is a Dynamic Storage Tiering attribute of a given volume in a
volume set of a multi-volume file system. This attribute is a character string, and
is known as a volume tag. A volume may have different tags, one of which could
be the placment class. The placement class tag makes a volume distinguishable
by DST.
Volume tags are organized as hierarchical name spaces in which the levels of the
hierarchy are separated by periods. By convention, the uppermost level in the
volume tag hierarchy denotes the Storage Foundation component or application
that uses a tag, and the second level denotes the tag’s purpose. DST recognizes
volume tags of the form vxfs.placement_class.class_name. The prefix vxfs
identifies a tag as being associated with VxFS. placement_class identifies the
tag as a file placement class used by DST. class_name represents the name of the
file placement class to which the tagged volume belongs. For example, a volume
with the tag vxfs.placement_class.tier1 belongs to placement class tier1.
Administrators use the vxvoladm command to associate tags with volumes.
See the vxadm(1M) manual page.
VxFS policy rules specify file placement in terms of placement classes rather than
in terms of individual volumes. All volumes that belong to a particular placement
class are interchangeable with respect to file creation and relocation operations.
Specifying file placement in terms of placement classes rather than in terms of
specific volumes simplifies the administration of multi-tier storage in the following
ways:
■ Adding or removing volumes does not require a file placement policy change.
If a volume with a tag value of vxfs.placement_class.tier2 is added to a file
system’s volume set, all policies that refer to tier2 immediately apply to the
newly added volume with no administrative action. Similarly, volumes can be
evacuated, that is, have data removed from them, and be removed from a file
system without a policy change. The active policy continues to apply to the
file system’s remaining volumes.
■ File placement policies are not specific to individual file systems. A file
placement policy can be assigned to any file system whose volume set includes
volumes tagged with the tag values (placement classes) named in the policy.
This property makes it possible for data centers with large numbers of servers
to define standard placement policies and apply them uniformly to all servers
with a single administrative action.
Administering placement policies
Tagging volumes as placement classes

The following example tags the vsavola volume as placement class tier1, vsavolb
as placement class tier2, vsavolc as placement class tier3, and vsavold as
placement class tier4 using the vxadm command.
To tag volumes
◆ Tag the volumes as placement classes:
# vxvoladm -g cfsdg settag vsavola vxfs.placement_class.tier1

# vxvoladm -g cfsdg settag vsavolb vxfs.placement_class.tier2
# vxvoladm -g cfsdg settag vsavolc vxfs.placement_class.tier3
# vxvoladm -g cfsdg settag vsavold vxfs.placement_class.tier4
Listing placement classes

Placement classes are listed using the vxvoladm listtag command.
See the vxvoladm(1M) manual page.
The following example lists all volume tags, including placment classes, set on a
volume vsavola in the diskgroup cfsdg.
To list placement classes
◆ List the volume tags, including placement classes:
# vxvoladm -g cfsdg listtag vsavola

A VxFS file placement policy document contains rules by which VxFS creates,
relocates, and deletes files, but the placement policy does not refer to specific file
systems or volumes. You can create a file system's active file placement policy by
assigning a placement policy document to the file system via the fsppadm command
or the GUI.
See the fsppadm(1M) manual page.
At most, one file placement policy can be assigned to a VxFS file system at any
time. A file system may have no file placement policy assigned to it, in which case
VxFS allocates space for new files according to its own internal algorithms.
In systems with Storage Foundation Management Server (SFMS) software installed,
file placement policy information is stored in the SFMS database. The SFMS
database contains both XML policy documents and lists of hosts and file systems
for which each document is the current active policy. When a policy document is
updated, SFMS can assign the updated document to all file systems whose current
active policies are based on that document. By default, SFMS does not update file
system active policies that have been created or modified locally, that is by the
hosts that control the placement policies' file systems. If a SFMS administrator
forces assignment of a placement policy to a file system, the file system's active
placement policy is overwritten and any local changes that had been made to the
placement policy are lost.
Assigning a placement policy

The following example uses the fsppadm assign command to assign the file
placement policy represented in the XML policy document /tmp/policy1.xml for
the file system at mount point /mnt1.
To assign a placement policy
◆ Assign a placement policy to a file system:
# fsppadm assign /mnt1 /tmp/policy1.xml
Unassigning a placement policy

The following example uses the fsppadm unassign command to unassign the
active file placement policy from the file system at mount point /mnt1.
To unassign a placement policy
◆ Unassign the placement policy from a file system:
# fsppadm unassign /mnt1
Analyzing the space impact of enforcing a placement policy

The following example uses the fsppadm analyze command to analyze the impact
if the enforce operation was performed on the mount point /mnt1. The command
builds the I/O temperature database if necessary.
To analyze the space impact of enforcing a placement policy
◆ Analyze the impact:
# fsppadm analyze -i /mnt1

Querying which files will be affected by enforcing a placement policy

The following example uses the fsppadm query command to generate a list of
files that will be affected by enforcing a placement policy. The command provides
details about where the files currently reside, to where the files will be relocated,
and which rule in the placement policy applies to the files.
To query which files will be affected by enforcing a placement policy
◆ Query the files:
# fsppadm query /mnt1/dir1/dir2 /mnt2 /mnt1/dir3
Enforcing a placement policy

Enforcing a placement policy for a file system requires that the policy be assigned
to the file system. You must assign a placement policy before it can be enforced.
Enforce operations are logged in a hidden file, .__fsppadm_enforce.log, in the
lost+found directory of the mount point. This log file contains details such as
files' previous locations, the files' new locations, and the reasons for the files'
relocations. The enforce operation creates the .__fsppadm_enforce.log file if
the file does not exist. The enforce operation appends the file if the file already
exists. The .__fsppadm_enforce.log file can be backed up or removed as with a
normal file.
The following example uses the fsppadm enforce command to enforce the file
placement policy for the file system at mount point /mnt1, and includes the access
time, modification time, and file size of the specified paths in the report,
/tmp/report.
File placement policy grammar
To enforce a placement policy

◆ Enforce a placement policy to a file system:
# fsppadm enforce -a -r /tmp/report /mnt1

Current Current Relocated Relocated
Class Volume Class Volume Rule File
tier3 dstvole tier3 dstvole a_to_z /mnt1/mds1/d1/file1
tier3 dstvole tier3 dstvole a_to_z /mnt1/mds1/d1/file2
tier3 dstvole tier3 dstvole a_to_z /mnt1/mds1/d1/d2/file3
tier3 dstvolf tier3 dstvolf a_to_z /mnt1/mds1/d1/d2/file4
.
.
.
Sweep path : /mnt1
Files moved : 42
KB moved : 1267
Tier Name Size (KB) Free Before (KB) Free After (KB)
tier4 524288 524256 524256
tier3 524288 522968 522968
tier2 524288 524256 524256
tier1 524288 502188 501227
Validating a placement policy

The following example uses the fsppadm validate command to validate the
placement policy policy.xml against all mounted file systems.
To validate a placement policy against all mounted file systems
◆ Validate the placement policy:
# fsppadm validate /tmp/policy.xml

VxFS allocates and relocates files within a multi-volume file system based on
properties in the file system metadata that pertains to the files. Placement
decisions may be based on file name, directory of residence, time of last access,
access frequency, file size, and ownership. An individual file system's criteria for
allocating and relocating files are expressed in the file system's file placement
policy.
A VxFS file placement policy defines the desired placement of sets of files on the
volumes of a VxFS multi-volume file system. A file placement policy specifies the
placement classes of volumes on which files should be created, and where and
under what conditions the files should be relocated to volumes in alternate
placement classes or deleted. You can create file placement policy documents,
which are XML text files, using either an XML or text editor, or a VxFS graphical
interface wizard.
The following output shows the overall structure of a placement policy:
<?xml version="1.0"?>
<!DOCTYPE PLACEMENT_POLICY [


<!ELEMENT PLACEMENT_POLICY (COMMENT?, RULE+)>


<!ATTLIST PLACEMENT_POLICY
Name CDATA #REQUIRED
Version (5.0) #REQUIRED
>


<!ELEMENT COMMENT (#PCDATA)>

<!ELEMENT RULE (COMMENT?, SELECT+, CREATE?, DELETE*, RELOCATE*)>


<!ATTLIST RULE
Name CDATA #REQUIRED
Flags (data) #REQUIRED
>

<!ELEMENT SELECT (COMMENT?, DIRECTORY*, PATTERN*, USER*, GROUP*)>

<!ATTLIST SELECT
Name CDATA #IMPLIED
>

<!ELEMENT DIRECTORY (#PCDATA)>


1. recursive
2. nonrecursive
If a given directory appears in more than one RULE,

all such DIRECTORY elements must all be recursive or
nonrecursive but can not be a combination. If no DIRECTORY
element is specified, all the files under the mount point
will be selected.
-->
<!ATTLIST DIRECTORY
Flags (recursive|nonrecursive) #REQUIRED
>

<!ELEMENT PATTERN (#PCDATA)>


<!ATTLIST PATTERN
Flags (recursive | nonrecursive) "nonrecursive"
>

<!ELEMENT USER (#PCDATA)>

<!ELEMENT GROUP (#PCDATA)>

<!ELEMENT CREATE (COMMENT?, ON)>

<!ATTLIST CREATE
Name CDATA #IMPLIED
Flags CDATA #IMPLIED
>

<!ELEMENT ON (DESTINATION*)>


<!ATTLIST ON
Name CDATA #IMPLIED
Flags (any) #IMPLIED
>

<!ELEMENT DESTINATION (CLASS?, PERCENT?, BALANCE_SIZE?)>


<!ATTLIST DESTINATION
Name CDATA #IMPLIED
Flags (disallow) #IMPLIED
>

<!ELEMENT CLASS (#PCDATA)>

<!ELEMENT PERCENT (#PCDATA)>

<!ELEMENT BALANCE_SIZE (#PCDATA)>


<!ATTLIST BALANCE_SIZE
Units (bytes|KB|MB|GB) #REQUIRED
>

<!ELEMENT DELETE (COMMENT?, FROM?, WHEN?)>

<!ATTLIST DELETE
Name CDATA #IMPLIED
Flags (none) #IMPLIED
>

<!ELEMENT RELOCATE (COMMENT?, FROM?, TO, WHEN?)>

<!ATTLIST RELOCATE
Name CDATA #IMPLIED
>

<!ELEMENT FROM (SOURCE+)>

<!ATTLIST FROM
Name CDATA #IMPLIED
>

<!ELEMENT SOURCE (CLASS, PERCENT?)>

<!ATTLIST SOURCE
Name CDATA #IMPLIED
>

<!ELEMENT TO (DESTINATION+)>

<!ATTLIST TO
Name CDATA #IMPLIED
>

<!ELEMENT WHEN (SIZE?, ACCAGE?, MODAGE?, IOTEMP?, ACCESSTEMP?)>

<!ATTLIST WHEN
Name CDATA #IMPLIED
>

<!ELEMENT SIZE (MIN?, MAX?)>


<!ATTLIST SIZE
Prefer (low|high) #IMPLIED
Units (bytes|KB|MB|GB) #REQUIRED
>

<!ELEMENT ACCAGE (MIN?, MAX?)>

1. low
2. high

1. hours
2. days
-->
<!ATTLIST ACCAGE
Units (hours|days) #REQUIRED
>

<!ELEMENT MODAGE (MIN?, MAX?)>

1. low
2. high

1. hours
2. days
-->
<!ATTLIST MODAGE
Units (hours|days) #REQUIRED
>

<!ELEMENT IOTEMP (MIN?, MAX?, PERIOD)>

1. low
2. high
-->

<!ATTLIST IOTEMP
Type (nrbytes|nwbytes|nrwbytes) #REQUIRED
>

<!ELEMENT ACCESSTEMP (MIN?, MAX?, PERIOD)>

1. low
2. high
-->

<!ATTLIST ACCESSTEMP
Type (nreads|nwrites|nrws) #REQUIRED
>


<!ELEMENT MIN (#PCDATA)>

1. gt for greater than
2. eq for equal to
3. gteq for greater than or equal to
-->
<!ATTLIST MIN
Flags (gt|eq|gteq) #REQUIRED
>


<!ELEMENT MAX (#PCDATA)>

1. lt for less than
2. lteq for less than or equal to
-->
File placement policy rules
<!ATTLIST MAX
Flags (lt|lteq) #REQUIRED
>


<!ELEMENT PERIOD (#PCDATA)>

<!ATTLIST PERIOD
Units (days) #REQUIRED
>
]>

A VxFS file placement policy consists of one or more rules. Each rule applies to
one or more files. The files to which a rule applies are designated in one or more
SELECT statements. A SELECT statement designates files according to one or
more of four properties: their names or naming patterns, the directories in which
they reside, their owners' user names, and their owners' group names.
A file may be designated by more than one rule. For example, if one rule designates
files in directory /dir, and another designates files owned by user1, a file in /dir
that is owned by user1 is designated by both rules. Only the rule that appears first
in the placement policy applies to the file; subsequent rules are ignored.
You can define placement policies that do not encompass the entire file system
name space. When a file that is not designated by any rule in its file system's
active placement policy is created, VxFS places the file according to its own internal
algorithms. To maintain full control over file placement, include a catchall rule
at the end of each placement policy document with a SELECT statement that
designates files by the naming pattern *. Such a rule designates all files that have
not been designated by the rules appearing earlier in the placement policy
document.
SELECT statement
The VxFS placement policy rule SELECT statement designates the collection of
files to which a rule applies.
The following XML snippet illustrates the general form of the SELECT statement:
<SELECT>
<DIRECTORY Flags="directory_flag_value">...value...
</DIRECTORY>
<PATTERN Flags="pattern_flag_value">...value...</PATTERN>
<USER>...value...</USER>
<GROUP>...value...</GROUP>
</SELECT>
A SELECT statement may designate files by using the following selection criteria:
<DIRECTORY> A full path name relative to the file system mount point. The
Flags=”directory_flag_value” XML attribute must have a value
of nonrecursive, denoting that only files in the specified directory
are designated, or a value of recursive, denoting that files in all
subdirectories of the specified directory are designated. The Flags
attribute is mandatory.
The <DIRECTORY> criterion is optional, and may be specified more

than once.
<PATTERN> Either an exact file name or a pattern using a single wildcard character
(*). For example, the pattern “abc*” denotes all files whose names begin
with “abc”. The pattern “abc.*” denotes all files whose names are
exactly "abc" followed by a period and any extension. The pattern
“*abc” denotes all files whose names end in “abc”, even if the name is
all or part of an extension. The pattern “*.abc” denotes files of any
name whose name extension (following the period) is “abc”. The
pattern “ab*c” denotes all files whose names start with “ab” and end
with “c”. The first "*" character is treated as a wildcard, while any
subsequent "*" characters are treated as literal text. The pattern cannot
contain "/".
The wildcard character matches any character, including ".", "?", and
"[", unlike using the wildcard in a shell.
The Flags=”pattern_flag_value” XML attribute is optional, and

if specified can only have a value of recursive. Specify
Flags="recursive" only if the pattern is a directory. If Flags is
not specified, the default attribute value is nonrecursive. If
Flags="recursive" is specified, the enclosing selection criteria
selects all files in any component directory that is anywhere below
the directory specified by <DIRECTORY> if the component directory
matches the pattern and either of the following is true:
■ <DIRECTORY> is specified and has the recursive flag.

■ <DIRECTORY> is not specified and the directory is anywhere in
the file system.
If the pattern contains the wildcard character (*), wildcard character

matching is performed.
The <PATTERN> criterion is optional, and may be specified more than
once. Only one value can be specified per <PATTERN> element.
<USER> User name of the file's owner. The user number cannot be specified
in place of the name.
The <USER> criterion is optional, and may be specified more than

once.
<GROUP> Group name of the file's owner. The group number cannot be specified
in place of the group name.
The <GROUP> criterion is optional, and may be specified more than

once.
One or more instances of any or all of the file selection criteria may be specified
within a single SELECT statement. If two or more selection criteria of different
types are specified in a single statement, a file must satisfy one criterion of each
type to be selected.
In the following example, only files that reside in either the ora/db or the
crash/dump directory, and whose owner is either user1 or user2 are selected for
possible action:
<SELECT>
<DIRECTORY Flags="nonrecursive">ora/db</DIRECTORY>
<DIRECTORY Flags="nonrecursive">crash/dump</DIRECTORY>
<USER>user1</USER>
<USER>user2</USER>
</SELECT>
A rule may include multiple SELECT statements. If a file satisfies the selection
criteria of one of the SELECT statements, it is eligible for action.
In the following example, any files owned by either user1 or user2, no matter in
which directories they reside, as well as all files in the ora/db or crash/dump
directories, no matter which users own them, are eligible for action:
<SELECT>
<DIRECTORY Flags="nonrecursive">ora/db</DIRECTORY>
<DIRECTORY Flags="nonrecursive">crash/dump</DIRECTORY>
</SELECT>
<SELECT>
<USER>user1</USER>
<USER>user2</USER>
</SELECT>
When VxFS creates new files, VxFS applies active placement policy rules in the
order of appearance in the active placement policy's XML source file. The first
rule in which a SELECT statement designates the file to be created determines
the file's placement; no later rules apply. Similarly, VxFS scans the active policy
rules on behalf of each file when relocating files, stopping the rules scan when it
reaches the first rule containing a SELECT statement that designates the file. This
behavior holds true even if the applicable rule results in no action. Take for
example a policy rule that indicates that .dat files inactive for 30 days should be
relocated, and a later rule indicates that .dat files larger than 10 megabytes should
be relocated. A 20 megabyte .dat file that has been inactive for 10 days will not
be relocated because the earlier rule applied. The later rule is never scanned.
A placement policy rule's action statements apply to all files designated by any
of the rule's SELECT statements. If an existing file is not designated by a SELECT
statement in any rule of a file system's active placement policy, then DST does
not relocate or delete the file. If an application creates a file that is not designated
by a SELECT statement in a rule of the file system's active policy, then VxFS places
the file according to its own internal algorithms. If this behavior is inappropriate,
the last rule in the policy document on which the file system's active placement
policy is based should specify <PATTERN>*</PATTERN> as the only selection
criterion in its SELECT statement, and a CREATE statement naming the desired
placement class for files not selected by other rules.
CREATE statement
A CREATE statement in a file placement policy rule specifies one or more
placement classes of volumes on which VxFS should allocate space for new files
to which the rule applies at the time the files are created. You can specify only
placement classes, not individual volume names, in a CREATE statement.
A file placement policy rule may contain at most one CREATE statement. If a rule
does not contain a CREATE statement, VxFS places files designated by the rule's
SELECT statements according to its internal algorithms. However, rules without
CREATE statements can be used to relocate or delete existing files that the rules'
SELECT statements designate.
The following XML snippet illustrates the general form of the CREATE statement:
<CREATE>
<ON Flags="...flag_value...">
<DESTINATION>
<CLASS>...placement_class_name...</CLASS>
<BALANCE_SIZE Units="units_specifier">...chunk_size...
</BALANCE_SIZE>
</DESTINATION>
<DESTINATION>...additional placement class specifications...
</DESTINATION>
</ON>
</CREATE>
A CREATE statement includes a single <ON> clause, in which one or more

<DESTINATION> XML elements specify placement classes for initial file allocation
in order of decreasing preference. VxFS allocates space for new files to which a
rule applies on a volume in the first class specified, if available space permits. If
space cannot be allocated on any volume in the first class, VxFS allocates space
on a volume in the second class specified if available space permits, and so forth.
If space cannot be allocated on any volume in any of the placement classes
specified, file creation fails with an ENOSPC error, even if adequate space is
available elsewhere in the file system's volume set. This situation can be
circumvented by specifying a Flags attribute with a value of “any” in the <ON>
clause. If <ON Flags=”any”> is specified in a CREATE statement, VxFS first
attempts to allocate space for new files to which the rule applies on the specified
placement classes. Failing that, VxFS resorts to its internal space allocation
algorithms, so file allocation does not fail unless there is no available space
any-where in the file system's volume set.
The Flags=”any” attribute differs from the catchall rule in that this attribute
applies only to files designated by the SELECT statement in the rule, which may
be less inclusive than the <PATTERN>*</PATTERN> file selection specification
of the catchall rule.
In addition to the placement class name specified in the <CLASS> sub-element,
a <DESTINATION> XML element may contain a <BALANCE_SIZE> sub-element.
Presence of a <BALANCE_SIZE> element indicates that space allocation should
be distributed across the volumes of the placement class in chunks of the indicated
size. For example, if a balance size of one megabyte is specified for a placement
class containing three volumes, VxFS allocates the first megabyte of space for a
new or extending file on the first (lowest indexed) volume in the class, the second
megabyte on the second volume, the third megabyte on the third volume, the
fourth megabyte on the first volume, and so forth. Using the Units attribute in
the <BALANCE_SIZE> XML tag, the balance size value may be specified in the
following units:
bytes Bytes
KB Kilobytes
MB Megabytes
GB Gigabytes
The <BALANCE_SIZE> element distributes the allocation of database files across

the volumes in a placement class. In principle, distributing the data in each file
across multiple volumes distributes the I/O load across the volumes as well.
The CREATE statement in the following example specifies that files to which the
rule applies should be created on the tier1 volume if space is available, and on one
of the tier2 volumes if not. If space allocation on tier1 and tier2 volumes is not
possible, file creation fails, even if space is available on tier3 volumes.
<CREATE>
<ON>
<DESTINATION>
<CLASS>tier1</CLASS>
</DESTINATION>
<DESTINATION>
<BALANCE_SIZE Units="MB">1</BALANCE_SIZE>
</DESTINATION>
</ON>
</CREATE>
The <BALANCE_SIZE> element with a value of one megabyte is specified for

allocations on tier2 volumes. For files allocated on tier2 volumes, the first megabyte
would be allocated on the first volume, the second on the second volume, and so
forth.
RELOCATE statement
The RELOCATE action statement of file placement policy rules specifies an action
that VxFS takes on designated files during periodic scans of the file system, and
the circumstances under which the actions should be taken. The fsppadm enforce
command is used to scan all or part of a file system for files that should be relocated
based on rules in the active placement policy at the time of the scan.
See the fsppadm(1M) manual page.
The fsppadm enforce scans file systems in path name order. For each file, VxFS
identifies the first applicable rule in the active placement policy, as determined
by the rules' SELECT statements. If the file resides on a volume specified in the
<FROM> clause of one of the rule's RELOCATE statements, and if the file meets
the criteria for relocation specified in the statement's <WHEN> clause, the file is
scheduled for relocation to a volume in the first placement class listed in the <TO>
clause that has space available for the file. The scan that results from issuing the
fsppadm enforce command runs to completion before any files are relocated.
The following XML snippet illustrates the general form of the RELOCATE
statement:
<RELOCATE>
<FROM>
<SOURCE>
</SOURCE>
<SOURCE>...additional placement class specifications...
</SOURCE>
</FROM>
<TO>
<DESTINATION>
<BALANCE_SIZE Units="units_specifier">
...chunk_size...
</BALANCE_SIZE>
</DESTINATION>
<DESTINATION>
...additional placement class specifications...
</DESTINATION>
</TO>
<WHEN>...relocation conditions...</WHEN>
</RELOCATE>
A RELOCATE statement contains the following clauses:
<FROM> An optional clause that contains a list of placement classes from whose
volumes designated files should be relocated if the files meet the
conditions specified in the <WHEN> clause. No priority is associated
with the ordering of placement classes listed in a <FROM> clause. If
a file to which the rule applies is located on a volume in any specified
placement class, the file is considered for relocation.
If a RELOCATE statement contains a <FROM> clause, VxFS only

considers files that reside on volumes in placement classes specified
in the clause for relocation. If no <FROM> clause is present, qualifying
files are relocated regardless of where the files reside.
<TO> Indicates the placement classes to which qualifying files should be

relocated. Unlike the source placement class list in a FROM clause,
placement classes in a <TO> clause are specified in priority order.
Files are relocated to volumes in the first specified placement class if
possible, to the second if not, and so forth.
The <TO> clause of the RELOCATE statement contains a list of

<DESTINATION> XML elements specifying placement classes to whose
volumes VxFS relocates qualifying files. Placement classes are
specified in priority order. VxFS relocates qualifying files to volumes
in the first placement class specified as long as space is available. A
<DESTINATION> element may contain an optional <BALANCE_SIZE>
modifier sub-element. The <BALANCE_SIZE> modifier indicates that
relocated files should be distributed across the volumes of the
destination placement class in chunks of the indicated size. For
example, if a balance size of one megabyte is specified for a placement
class containing three volumes, VxFS relocates the first megabyte the
file to the first (lowest indexed) volume in the class, the second
megabyte to the second volume, the third megabyte to the third
volume, the fourth megabyte to the first volume, and so forth. Using
the Units attribute in the <BALANCE_SIZE> XML tag, the chunk value
may be specified in the balance size value may be specified in bytes
(Units=”bytes”), kilobytes (Units=”KB”), megabytes (Units=”MB”), or
gigabytes (Units=”GB”).
The <BALANCE_SIZE> element distributes the allocation of database

files across the volumes in a placement class. In principle, distributing
the data in each file across multiple volumes distributes the I/O load
across the volumes as well.
<WHEN> An optional clause that indicates the conditions under which files to
which the rule applies should be relocated. Files that have been
unaccessed or unmodified for a specified period, reached a certain
size, or reached a specific I/O temperature or access temperature level
may be relocated. If a RELOCATE statement does not contain a
<WHEN> clause, files to which the rule applies are relocated
unconditionally.
A <WHEN> clause may be included in a RELOCATE statement to

specify that files should be relocated only if any or all of four types of
criteria are met. Files can be specified for relocation if they satisfy
one or more criteria.
The following are the criteria that can be specified for the <WHEN> clause:
<ACCAGE> This criterion is met when files are inactive for a designated period
or during a designated period relative to the time at which the
fsppadm enforce command was issued.
<MODAGE> This criterion is met when files are unmodified for a designated
period or during a designated period relative to the time at which
the fsppadm enforce command was issued.
<SIZE> This criterion is met when files exceed or drop below a designated
size or fall within a designated size range.
<IOTEMP> This criterion is met when files exceed or drop below a designated
I/O temperature, or fall with in a designated I/O temperature range.
A file's I/O temperature is a measure of the I/O activity against it
during the period designated by the <PERIOD> element prior to
the time at which the fsppadm enforce command was issued.
See “Calculating I/O temperature and access temperature”

on page 174.
<ACCESSTEMP> This criterion is met when files exceed or drop below a specified
average access temperature, or fall within a specified access
temperature range. A file's access temperature is similar to its I/O
temperature, except that access temperature is computed using
the number of I/O requests to the file, rather than the number of
bytes transferred.
The following XML snippet illustrates the general form of the <WHEN> clause in
a RELOCATE statement:
<WHEN>
<ACCAGE Units="...units_value...">
<MIN Flags="...comparison_operator...">
...min_access_age...</MIN>
<MAX Flags="...comparison_operator...">
...max_access_age...</MAX>
</ACCAGE>
<MODAGE Units="...units_value...">
...min_modification_age...</MIN>
...max_modification_age...</MAX>
</MODAGE>
<SIZE " Units="...units_value...">
...min_size...</MIN>
...max_size...</MAX>
</SIZE>
<IOTEMP Type="...read_write_preference...">
...min_I/O_temperature...</MIN>
...max_I/O_temperature...</MAX>
<PERIOD>...days_of_interest...</PERIOD>
</IOTEMP>
<ACCESSTEMP Type="...read_write_preference...">
...min_access_temperature...</MIN>
...max_access_temperature...</MAX>
<PERIOD>...days_of_interest...</PERIOD>
</ACCESSTEMP>
</WHEN>
The access age (<ACCAGE>) element refers to the amount of time since a file was
last accessed. VxFS computes access age by subtracting a file's time of last access,
atime, from the time when the fsppadm enforce command was issued. The <MIN>
and <MAX> XML elements in an <ACCAGE> clause, denote the minimum and
maximum access age thresholds for relocation, respectively. These elements are
optional, but at least one must be included. Using the Units XML attribute, the
<MIN> and <MAX> elements may be specified in the following units:
hours Hours
days Days. A day is considered to be 24 hours prior to the time that the
fsppadm enforce command was issued.
Both the <MIN> and <MAX> elements require Flags attributes to direct their
operation.
For <MIN>, the following Flags attributes values may be specified:
gt The time of last access must be greater than the specified interval.
eq The time of last access must be equal to the specified interval.
gteq The time of last access must be greater than or equal to the specified
interval.
For <MAX>, the following Flags attributes values may be specified.
lt The time of last access must be less than the specified interval.
lteq The time of last access must be less than or equal to the specified
interval.
Including a <MIN> element in a <WHEN> clause causes VxFS to relocate files to

which the rule applies that have been inactive for longer than the specified interval.
Such a rule would typically be used to relocate inactive files to less expensive
storage tiers. Conversely, including <MAX> causes files accessed within the
specified interval to be relocated. It would typically be used to move inactive files
against which activity had recommenced to higher performance or more reliable
storage. Including both <MIN> and <MAX> causes VxFS to relocate files whose
access age lies between the two.
The modification age relocation criterion, <MODAGE>, is similar to access age,
except that files' POSIX mtime values are used in computations. You would
typically specify the <MODAGE> criterion to cause relocation of recently modified
files to higher performance or more reliable storage tiers in anticipation that the
files would be accessed recurrently in the near future.
The file size relocation criterion, <SIZE>, causes files to be relocated if the files
are larger or smaller than the values specified in the <MIN> and <MAX> relocation
criteria, respectively, at the time that the fsppadm enforce command was issued.
Specifying both criteria causes VxFS to schedule relocation for files whose sizes
lie between the two. Using the Units attribute, threshold file sizes may be specified
in the following units:
bytes Bytes
KB Kilobytes
MB Megabytes
GB Gigabytes
Specifying the I/O temperature relocation criterion

The I/O temperature relocation criterion, <IOTEMP>, causes files to be relocated
if their I/O temperatures rise above or drop below specified values over a specified
period immediately prior to the time at which the fsppadm enforce command
was issued. A file's I/O temperature is a measure of the read, write, or total I/O
activity against it normalized to the file's size. Higher I/O temperatures indicate
higher levels of application activity; lower temperatures indicate lower levels.
VxFS computes a file's I/O temperature by dividing the number of bytes transferred
to or from it (read, written, or both) during the specified period by its size at the
time that the fsppadm enforce command was issued.
See “Calculating I/O temperature and access temperature” on page 174.
As with the other file relocation criteria, <IOTEMP> may be specified with a lower
threshold by using the <MIN> element, an upper threshold by using the <MAX>
element, or as a range by using both. However, I/O temperature is dimensionless
and therefore has no specification for units.
VxFS computes files' I/O temperatures over the period between the time when
the fsppadm enforce command was issued and the number of days in the past
specified in the <PERIOD> element, where a day is a 24 hour period. For example,
if the fsppadm enforce command was issued at 2 PM on Wednesday, and a
<PERIOD> value of 2 was specified, VxFS looks at file I/O activity for the period
between 2 PM on Monday and 2 PM on Wednesday. The number of days specified
in the <PERIOD> element should not exceed one or two weeks due to the disk
space used by the File Change Log (FCL) file.
I/O temperature is a softer measure of I/O activity than access age. With access
age, a single access to a file resets the file's atime to the current time. In contrast,
a file's I/O temperature decreases gradually as time passes without the file being
accessed, and increases gradually as the file is accessed periodically. For example,
if a new 10 megabyte file is read completely five times on Monday and fsppadm
enforce runs at midnight, the file's two-day I/O temperature will be five and its
access age in days will be zero. If the file is read once on Tuesday, the file's access
age in days at midnight will be zero, and its two-day I/O temperature will have
dropped to three. If the file is read once on Wednesday, the file's access age at
midnight will still be zero, but its two-day I/O temperature will have dropped to
one, as the influence of Monday's I/O will have disappeared.
If the intention of a file placement policy is to keep files in place, such as on top-tier
storage devices, as long as the files are being accessed at all, then access age is
the more appropriate relocation criterion. However, if the intention is to relocate
files as the I/O load on them decreases, then I/O temperature is more appropriate.
The case for upward relocation is similar. If files that have been relocated to
lower-tier storage devices due to infrequent access experience renewed application
activity, then it may be appropriate to relocate those files to top-tier devices. A
policy rule that uses access age with a low <MAX> value, that is, the interval
between fsppadm enforce runs, as a relocation criterion will cause files to be
relocated that have been accessed even once during the interval. Conversely, a
policy that uses I/O temperature with a <MIN> value will only relocate files that
have experienced a sustained level of activity over the period of interest.
RELOCATE statement examples

The following example illustrates an unconditional relocation statement, which
is the simplest form of the RELOCATE policy rule statement:
<RELOCATE>
<FROM>
<SOURCE>
</SOURCE>
</FROM>
<TO>
<DESTINATION>
</DESTINATION>
</TO>
</RELOCATE>
The files designated by the rule's SELECT statement that reside on volumes in
placement class tier1 at the time the fsppadm enforce command executes would
be unconditionally relocated to volumes in placement class tier2 as long as space
permitted. This type of rule might be used, for example, with applications that
create and access new files but seldom access existing files once they have been
processed. A CREATE statement would specify creation on tier1 volumes, which
are presumably high performance or high availability, or both. Each instantiation
of fsppadm enforce would relocate files created since the last run to tier2 volumes.
The following example illustrates a more comprehensive form of the RELOCATE
statement that uses access age as the criterion for relocating files from tier1
volumes to tier2 volumes. This rule is designed to maintain free space on tier1
volumes by relocating inactive files to tier2 volumes:
<RELOCATE>
<FROM>
<SOURCE>
</SOURCE>
</FROM>
<TO>
<DESTINATION>
</DESTINATION>
</TO>
<WHEN>
<SIZE Units="MB">
<MIN Flags="gt">1</MIN>
<MAX Flags="lt">1000</MAX>
</SIZE>
<ACCAGE Units="days">
</ACCAGE>
</WHEN>
</RELOCATE>
Files designated by the rule's SELECT statement are relocated from tier1 volumes
to tier2 volumes if they are between 1 MB and 1000 MB in size and have not been
accessed for 30 days. VxFS relocates qualifying files in the order in which it
encounters them as it scans the file system's directory tree. VxFS stops scheduling
qualifying files for relocation when when it calculates that already-scheduled
relocations would result in tier2 volumes being fully occupied.
The following example illustrates a possible companion rule that relocates files
from tier2 volumes to tier1 ones based on their I/O temperatures. This rule might
be used to return files that had been relocated to tier2 volumes due to inactivity
to tier1 volumes when application activity against them increases. Using I/O
temperature rather than access age as the relocation criterion reduces the chance
of relocating files that are not actually being used frequently by applications. This
rule does not cause files to be relocated unless there is sustained activity against
them over the most recent two-day period.
<RELOCATE>
<FROM>
<SOURCE>
</SOURCE>
</FROM>
<TO>
<DESTINATION>
</DESTINATION>
</TO>
<WHEN>
<IOTEMP Type="nrbytes">
<PERIOD>2</PERIOD>
</IOTEMP>
</WHEN>
</RELOCATE>
This rule relocates files that reside on tier2 volumes to tier1 volumes if their I/O
temperatures are above 5 for the two day period immediately preceding the issuing
of the fsppadm enforce command. VxFS relocates qualifying files in the order in
which it encounters them during its file system directory tree scan. When tier1
volumes are fully occupied, VxFS stops scheduling qualifying files for relocation.
VxFS file placement policies are able to control file placement across any number
of placement classes. The following example illustrates a rule for relocating files
with low I/O temperatures from tier1 volumes to tier2 volumes, and to tier3
volumes when tier2 volumes are fully occupied:
<RELOCATE>
<FROM>
<SOURCE>
</SOURCE>
</FROM>
<TO>
<DESTINATION>
</DESTINATION>
<DESTINATION>
</DESTINATION>
</TO>
<WHEN>
<PERIOD>3</PERIOD>
</IOTEMP>
</WHEN>
</RELOCATE>
This rule relocates files whose 3-day I/O temperatures are less than 4 and which
reside on tier1 volumes. When VxFS calculates that already-relocated files would
result in tier2 volumes being fully occupied, VxFS relocates qualifying files to
tier3 volumes instead. VxFS relocates qualifying files as it encounters them in its
scan of the file system directory tree.
The <FROM> clause in the RELOCATE statement is optional. If the clause is not
present, VxFS evaluates files designated by the rule's SELECT statement for
relocation no matter which volumes they reside on when the fsppadm enforce
command is issued. The following example illustrates a fragment of a policy rule
that relocates files according to their sizes, no matter where they reside when the
fsppadm enforce command is issued:
<RELOCATE>
<TO>
<DESTINATION>
</DESTINATION>
</TO>
<WHEN>
<SIZE Units="MB">
</SIZE>
</WHEN>
</RELOCATE>
<RELOCATE>
<TO>
<DESTINATION>
</DESTINATION>
</TO>
<WHEN>
<SIZE Units="MB">
<MIN Flags="gteq">10</MIN>
</SIZE>
</WHEN>
</RELOCATE>
<RELOCATE>
<TO>
<DESTINATION>
</DESTINATION>
</TO>
<WHEN>
<SIZE Units="MB">
<MIN Flags="gteq">100</MIN>
</SIZE>
</WHEN>
</RELOCATE>
This rule relocates files smaller than 10 megabytes to tier1 volumes, files between
10 and 100 megabytes to tier2 volumes, and files larger than 100 megabytes to
tier3 volumes. VxFS relocates all qualifying files that do not already reside on
volumes in their DESTINATION placement classes when the fsppadm enforce
command is issued.
DELETE statement
The DELETE file placement policy rule statement is very similar to the RELOCATE
statement in both form and function, lacking only the <TO> clause. File placement
policy-based deletion may be thought of as relocation with a fixed destination.
Note: Use DELETE statements with caution.
The following XML snippet illustrates the general form of the DELETE statement:
<DELETE>
<FROM>
<SOURCE>
</SOURCE>
<SOURCE>
...additional placement class specifications...
</SOURCE>
</FROM>
<WHEN>...relocation conditions...</WHEN>
</DELETE>
A DELETE statement contains the following clauses:
<FROM> An optional clause that contains a list of placement classes from whose
volumes designated files should be deleted if the files meet the
conditions specified in the <WHEN> clause. No priority is associated
with the ordering of placement classes in a <FROM> clause. If a file
to which the rule applies is located on a volume in any specified
placement class, the file is deleted. If a DELETE statement does not
contain a <FROM> clause, VxFS deletes qualifying files no matter on
which of a file system's volumes the files reside.
<WHEN> An optional clause specifying the conditions under which files to which
the rule applies should be deleted. The form of the <WHEN> clause
in a DELETE statement is identical to that of the <WHEN> clause in
a RELOCATE statement. If a DELETE statement does not contain a
<WHEN> clause, files designated by the rule's SELECT statement, and
the <FROM> clause if it is present, are deleted unconditionally.
DELETE statement examples

The following example illustrates the use of the DELETE statement:
Calculating I/O temperature and access temperature
<DELETE>
<FROM>
<SOURCE>
</SOURCE>
</FROM>
</DELETE>
<DELETE>
<FROM>
<SOURCE>
</SOURCE>
</FROM>
<WHEN>
</ACCAGE>
</WHEN>
</DELETE>
The first DELETE statement unconditionally deletes files designated by the rule's
SELECT statement that reside on tier3 volumes when the fsppadm enforce
command is issued. The absence of a <WHEN> clause in the DELETE statement
indicates that deletion of designated files is unconditional.
The second DELETE statement deletes files to which the rule applies that reside
on tier2 volumes when the fsppadm enforce command is issued and that have not
been accessed for the past 120 days.

An important application of VxFS Dynamic Storage Tiering is automating the
relocation of inactive files to lower cost storage. If a file has not been accessed for
the period of time specified in the <ACCAGE> element, a scan of the file system
should schedule the file for relocation to a lower tier of storage. But, time since
last access is inadequate as the only criterion for activity-based relocation for the
following reasons:
■ Access age is a binary measure. The time since last access of a file is computed
by subtracting the time at which the fsppadm enforce command is issued
from the POSIX atime in the file's metadata. If a file is opened the day before
the fsppadm enforce command, its time since last access is one day, even
though it may have been inactive for the month preceding. If the intent of a
policy rule is to relocate inactive files to lower tier volumes, it will perform
badly against files that happen to be accessed, however casually, within the
interval defined by the value of the <ACCAGE> pa-rameter.
■ Access age is a poor indicator of resumption of significant activity. Using
ACCAGE, the time since last access, as a criterion for relocating inactive files
to lower tier volumes may fail to schedule some relocations that should be
performed, but at least this method results in less relocat--ion activity than
necessary. Using ACCAGE as a criterion for relocating previously inactive files
that have become active is worse, because this method is likely to schedule
relocation activity that is not warranted. If a policy rule's intent is to cause
files that have experienced I/O activity in the recent past to be relocated to
higher performing, perhaps more failure tolerant storage, ACCAGE is too
coarse a filter. For example, in a rule specifying that files on tier2 volumes
that have been accessed within the last three days should be relocated to tier1
volumes, no distinction is made between a file that was browsed by a single
user and a file that actually was used intensively by applications.
DST implements the concept of I/O temperature and access temperature to
overcome these deficiencies. A file's I/O temperature is equal to the number of
bytes transferred to or from it over a specified period of time divided by the size
of the file. For example, if a file occupies one megabyte of storage at the time of
an fsppadm enforce operation and the data in the file has been completely read
or written 15 times within the last three days, VxFS calculates its 3-day average
I/O temperature to be 5 (15 MB of I/O ÷ 1 MB file size ÷ 3 days).
Similarly, a file's average access temperature is the number of read or write
requests made to it over a specified number of 24-hour periods divided by the
number of periods. Unlike I/O temperature, access temperature is unrelated to
file size. A large file to which 20 I/O requests are made over a 2-day period has
the same average access temperature as a small file accessed 20 times over a 2-day
period.
If a file system's active placement policy includes any <IOTEMP> or
<ACCESSTEMP> clauses, VxFS begins policy enforcement by using information
in the file system's FCL file to calculate average I/O activity against all files in the
file system during the longest <PERIOD> specified in the policy. Shorter specified
periods are ignored. VxFS uses these calculations to qualify files for I/O
temperature-based relocation and deletion.
Note: If FCL is turned off, I/O temperature-based relocation will not be accurate.
When you invoke the fsppadm enforce command, the command displays a
warning if the FCL is turned off.
As its name implies, the File Change Log records information about changes made
to files in a VxFS file system. In addition to recording creations, deletions,
extensions, the FCL periodically captures the cumulative amount of I/O activity
(number of bytes read and written) on a file-by-file basis. File I/O activity is
recorded in the FCL each time a file is opened or closed, as well as at timed intervals
to capture information about files that remain open for long periods.
If a file system's active file placement policy contains <IOTEMP> clauses, execution
of the fsppadm enforce command begins with a scan of the FCL to extract I/O
activity information over the period of interest for the policy. The period of interest
is the interval between the time at which the fsppadm enforce command was
issued and that time minus the largest interval value specified in any <PERIOD>
element in the active policy.
For files with I/O activity during the largest interval, VxFS computes an
approximation of the amount of read, write, and total data transfer (the sum of
the two) activity by subtracting the I/O levels in the oldest FCL record that pertains
to the file from those in the newest. It then computes each file's I/O temperature
by dividing its I/O activity by its size at Tscan. Dividing by file size is an implicit
acknowledgement that relocating larger files consumes more I/O resources than
relocating smaller ones. Using this algorithm requires that larger files must have
more activity against them in order to reach a given I/O temperature, and thereby
justify the resource cost of relocation.
While this computation is an approximation in several ways, it represents an easy
to compute, and more importantly, unbiased estimate of relative recent I/O activity
upon which reasonable relocation decisions can be based.
File relocation and deletion decisions can be based on read, write, or total I/O
activity.
The following XML snippet illustrates the use of IOTEMP in a policy rule to specify
relocation of low activity files from tier1 volumes to tier2 volumes:
<RELOCATE>
<FROM>
<SOURCE>
</SOURCE>
</FROM>
<TO>
<DESTINATION>
</DESTINATION>
</TO>
<WHEN>
<PERIOD Units="days">4</PERIOD>
</IOTEMP>
</WHEN>
</RELOCATE>
This snippet specifies that files to which the rule applies should be relocated from
tier1 volumes to tier2 volumes if their I/O temperatures fall below 3 over a period
of 4 days. The Type=”nrbytes” XML attribute specifies that total data transfer
activity, which is the the sum of bytes read and bytes written, should be used in
the computation. For example, a 50 megabyte file that experienced less than 150
megabytes of data transfer over the 4-day period immediately preceding the
fsppadm enforce scan would be a candidate for relocation. VxFS considers files
that experience no activity over the period of interest to have an I/O temperature
of zero. VxFS relocates qualifying files in the order in which it encounters the
files in its scan of the file system directory tree.
Using I/O temperature or access temperature rather than a binary indication of
activity, such as the POSIX atime or mtime, minimizes the chance of not relocating
files that were only accessed occasionally during the period of interest. A large
file that has had only a few bytes transferred to or from it would have a low I/O
temperature, and would therefore be a candidate for relocation to tier2 volumes,
even if the activity was very recent.
But, the greater value of I/O temperature or access temperature as a file relocation
criterion lies in upward relocation: detecting increasing levels of I/O activity
against files that had previously been relocated to lower tiers in a storage hierarchy
due to inactivity or low temperatures, and relocating them to higher tiers in the
storage hierarchy.
The following XML snippet illustrates relocating files from tier2 volumes to tier1
when the activity level against them increases.
<RELOCATE>
<FROM>
<SOURCE>
</SOURCE>
</FROM>
<TO>
<DESTINATION>
</DESTINATION>
</TO>
Multiple criteria in file placement policy rule statements
<WHEN>
<MAX Flags="gt">5</MAX>
<PERIOD Units="days">2</PERIOD>
</IOTEMP>
</WHEN>
</RELOCATE>
The <RELOCATE> statement specifies that files on tier2 volumes whose I/O
temperature as calculated using the number of bytes read is above 5 over a 2-day
period are to be relocated to tier1 volumes. Bytes written to the file during the
period of interest are not part of this calculation.
Using I/O temperature rather than a binary indicator of activity as a criterion for
file relocation gives administrators a granular level of control over automated
file relocation that can be used to attune policies to application requirements. For
example, specifying a large value in the <PERIOD> element of an upward relocation
statement prevents files from being relocated unless I/O activity against them is
sustained. Alternatively, specifying a high temperature and a short period tends
to relocate files based on short-term intensity of I/O activity against them.
I/O temperature and access temperature utilize the sqlite3 database for building
a temporary table indexed on an inode. This temporary table is used to filter files
based on I/O temperature and access temperature. The temporary table is stored
in the database file .__fsppadm_fcliotemp.db, which resides in the lost+found
directory of the mount point.
Multiple criteria in file placement policy rule

statements
In certain cases, file placement policy rule statements may contain multiple clauses
that affect their behavior. In general, when a rule statement contains multiple
clauses of a given type, all clauses must be satisfied in order for the statement to
be effective. There are four cases of note in which multiple clauses may be used.
Multiple file selection criteria in SELECT statement clauses

Within a single SELECT statement, all the selection criteria clauses of a single
type are treated as a selection list. A file need only satisfy a single criterion of a
given type to be designated.
In the following example, files in any of the db/datafiles, db/indexes, and
db/logs directories, all relative to the file system mount point, would be selected:
<SELECT>
<DIRECTORY Flags="nonrecursive">db/datafiles</DIRECTORY>
<DIRECTORY Flags="nonrecursive">db/indexes</DIRECTORY>
<DIRECTORY Flags="nonrecursive">db/logs</DIRECTORY>
</SELECT>
This example is in direct contrast to the treatment of selection criteria clauses of

different types. When a SELECT statement includes multiple types of file selection
criteria, a file must satisfy one criterion of each type in order for the rule's action
statements to apply.
In the following example, a file must reside in one of db/datafiles, db/indexes, or
db/logs and be owned by one of DBA_Manager, MFG_DBA, or HR_DBA to be
designated for possible action:
<SELECT>
<USER>DBA_Manager</USER>
<USER>MFG_DBA</USER>
<USER>HR_DBA</USER>
</SELECT>
If a rule includes multiple SELECT statements, a file need only satisfy one of them
to be selected for action. This property can be used to specify alternative conditions
for file selection.
In the following example, a file need only reside in one of db/datafiles,
db/indexes, or db/logs or be owned by one of DBA_Manager, MFG_DBA, or
HR_DBA to be designated for possible action:
<SELECT>
</SELECT>
<SELECT>
<USER>DBA_Manager</USER>
<USER>MFG_DBA</USER>
<USER>HR_DBA</USER>
</SELECT>
Multiple placement classes in <ON> clauses of CREATE statements

and in <TO> clauses of RELOCATE statements
Both the <ON> clause of the CREATE statement and the <TO> clause of the
RELOCATE statement can specify priority ordered lists of placement classes using
multiple <DESTINATION> XML elements. VxFS uses a volume in the first
placement class in a list for the designated purpose of file creation or relocation,
if possible. If no volume in the first listed class has sufficient free space or if the
file system's volume set does not contain any volumes with that placement class,
VxFS uses a volume in the second listed class if possible. If no volume in the second
listed class can be used, a volume in the third listed class is used if possible, and
so forth.
The following example illustrates of three placement classes specified in the <ON>
clause of a CREATE statement:
<CREATE>
<ON>
<DESTINATION>
</DESTINATION>
<DESTINATION>
</DESTINATION>
<DESTINATION>
</DESTINATION>
</ON>
</CREATE>
In this statement, VxFS would allocate space for newly created files designated
by the rule's SELECT statement on tier1 volumes if space was available. If no tier1
volume had sufficient free space, VxFS would attempt to allocate space on a tier2
volume. If no tier2 volume had sufficient free space, VxFS would attempt allocation
on a tier3 volume. If sufficient space could not be allocated on a volume in any of
the three specified placement classes, allocation would fail with an ENOSPC error,
even if the file system's volume set included volumes in other placement classes
that did have sufficient space.
The <TO> clause in the RELOCATE statement behaves similarly. VxFS relocates
qualifying files to volumes in the first placement class specified if possible, to
volumes in the second specified class if not, and so forth. If none of the destination
criteria can be met, such as if all specified classes are fully occupied, qualifying
files are not relocated, but no error is signaled in this case.
File placement policy rule and statement ordering
Multiple placement classes in <FROM> clauses of RELOCATE and

DELETE statements
The <FROM> clause in RELOCATE and DELETE statements can include multiple
source placement classes. However, unlike the <ON> and <TO> clauses, no order
or priority is implied in <FROM> clauses. If a qualifying file resides on a volume
in any of the placement classes specified in a <FROM> clause, it is relocated or
deleted regardless of the position of its placement class in the <FROM> clause
list of classes.
Multiple conditions in <WHEN> clauses of RELOCATE and DELETE

statements
The <WHEN> clause in RELOCATE and DELETE statements may include multiple
relocation criteria. Any or all of <ACCAGE>, <MODAGE>, <SIZE>, and <IOTEMP>
can be specified. When multiple conditions are specified, all must be satisfied in
order for a selected file to qualify for relocation or deletion.
In the following example, a selected file would have to be both inactive, that is,
not accessed, for more than 30 days and larger than 100 megabytes to be eligible
for relocation or deletion:
<WHEN>
</ACCAGE>
<SIZE Units="MB">
</SIZE>
</WHEN>
You cannot write rules to relocate or delete a single designated set of files if the
files meet one of two or more relocation or deletion criteria.

You can use the Dynamic Storage Tiering graphical user interface (GUI) to create
any of four types of file placement policy documents. Alternatively, you can use
a text editor or XML editor to create XML policy documents directly. The GUI
places policy rule statements in the correct order to achieve the desired behavior.
If you use a text editor, it is your responsibility to order policy rules and the
statements in them so that the desired behavior results.
The rules that comprise a placement policy may occur in any order, but during
both file allocation and fsppadm enforce relocation scans, the first rule in which
a file is designated by a SELECT statement is the only rule against which that file
is evaluated. Thus, rules whose purpose is to supersede a generally applicable
behavior for a special class of files should precede the general rules in a file
placement policy document.
The following XML snippet illustrates faulty rule placement with potentially
unintended consequences:
<?xml version="1.0"?>
<!DOCTYPE FILE_PLACEMENT_POLICY SYSTEM "placement.dtd">
<FILE_PLACEMENT_POLICY Version="5.0">
<RULE Name="GeneralRule">
<SELECT>
<PATTERN>*</PATTERN>
</SELECT>
<CREATE>
<ON>
<DESTINATION>
</DESTINATION>
</ON>
</CREATE>
...other statements...
</RULE>
<RULE Name="DatabaseRule">
<SELECT>
<PATTERN>*.db</PATTERN>
</SELECT>
<CREATE>
<ON>
<DESTINATION>
</DESTINATION>
</ON>
</CREATE>
...other statements...
</RULE>
</FILE_PLACEMENT_POLICY>
The GeneralRule rule specifies that all files created in the file system, designated
by <PATTERN>*</PATTERN>, should be created on tier2 volumes. The
DatabaseRule rule specifies that files whose names include an extension of .db
should be created on tier1 volumes. The GeneralRule rule applies to any file created
in the file system, including those with a naming pattern of *.db, so the
DatabaseRule rule will never apply to any file. This fault can be remedied by
exchanging the order of the two rules. If the DatabaseRule rule occurs first in the
policy document, VxFS encounters it first when determining where to new place
files whose names follow the pattern *.db, and correctly allocates space for them
on tier1 volumes. For files to which the DatabaseRule rule does not apply, VxFS
continues scanning the policy and allocates space according to the specification
in the CREATE statement of the GeneralRule rule.
A similar consideration applies to statements within a placement policy rule. VxFS
processes these statements in order, and stops processing on behalf of a file when
it encounters a statement that pertains to the file. This can result in unintended
behavior.
The following XML snippet illustrates a RELOCATE statement and a DELETE
statement in a rule that is intended to relocate if the files have not been accessed
in 30 days, and delete the files if they have not been accessed in 90 days:
<RELOCATE>
<TO>
<DESTINATION>
</DESTINATION>
</TO>
<WHEN>
</ACCAGE>
</WHEN>
</RELOCATE>
<DELETE>
<WHEN>
</ACCAGE>
</WHEN>
</DELETE>
As written with the RELOCATE statement preceding the DELETE statement, files
will never be deleted, because the <WHEN> clause in the RELOCATE statement
applies to all selected files that have not been accessed for at least 30 days. This
includes those that have not been accessed for 90 days. VxFS ceases to process a
file against a placement policy when it identifies a statement that applies to that
file, so the DELETE statement would never occur. This example illustrates the
File placement policies and extending files
general point that RELOCATE and DELETE statements that specify less inclusive
criteria should precede statements that specify more inclusive criteria in a file
placement policy document. The GUI automatically produce the correct statement
order for the policies it creates.
File placement policies and extending files

In a VxFS file system with an active file placement policy, the placement class on
whose volume a file resides is part of its metadata, and is attached when it is
created and updated when it is relocated. When an application extends a file, VxFS
allocates the incremental space on the volume occupied by the file if possible. If
not possible, VxFS allocates the space on another volume in the same placement
class. For example, if a file is created on a tier1 volume and later relocated to a
tier2 volume, extensions to the file that occur before the relocation have space
allocated on a tier1 volume, while those occurring after to the relocation have
their space allocated on tier2 volumes. When a file is relocated, all of its allocated
space, including the space acquired by extension, is relocated to tier2 volumes in
this case.
Chapter 11
Quick I/O for Databases
■ About Quick I/O
■ About Quick I/O functionality and performance
■ About using Veritas File System files as raw character devices
■ About creating a Quick I/O file using qiomkfile
■ Accessing regular VxFS files through symbolic links
■ Using Quick I/O with Oracle databases
■ Using Quick I/O with Sybase databases
■ Enabling and disabling Quick I/O
■ About Cached Quick I/O for databases
■ About Quick I/O statistics
■ Increasing database performance using Quick I/O
About Quick I/O

Quick I/O for Databases (referred to as Quick I/O) allows applications to access
preallocated VxFS files as raw character devices. This provides the administrative
benefits of running databases on file systems without the performance degradation
usually associated with databases created on file systems.
186 Quick I/O for Databases
About Quick I/O functionality and performance
Quick I/O is part of the VRTSvxfs package, but is available for use only with other
Symantec products.
See the Veritas Storage Foundation Release Notes.
About Quick I/O functionality and performance

Many database administrators (DBAs) create databases on file systems because
file systems make common administrative tasks, such as moving, copying, and
backing up, much simpler. However, putting databases on file systems significantly
reduces database performance. By using Quick I/O, you can retain the advantages
of having databases on file systems without performance degradation.
Quick I/O uses a special naming convention to allow database applications to
access regular files as raw character devices.
Quick I/O provides higher database performance in the following ways:
■ Supporting kernel asynchronous I/O
■ Supporting direct I/O
■ Avoiding kernel write locks
■ Avoiding double buffering
About asynchronous I/O kernel support
About direct I/O support

I/O on files using read() and write() system calls typically results in data being
copied twice: once between user and kernel space, and later between kernel space
and disk. In contrast, I/O on raw devices is direct. That is, data is copied directly
between user space and disk, saving one level of copying. As with I/O on raw
devices, Quick I/O avoids the extra copying.
About Kernel write locks avoidance

When database I/O is performed via the write() system call, each system call
acquires and releases a write lock inside the kernel. This lock prevents
simultaneous write operations on the same file. Because database systems usually
implement their own locks for managing concurrent access to files, write locks
unnecessarily serialize I/O operations. Quick I/O bypasses file system locking and
lets the database server control data access.
Quick I/O for Databases 187
About using Veritas File System files as raw character devices
About double buffering avoidance

Most database servers implement their own buffer cache and do not need the
system buffer cache. Thus, the memory used by the system buffer cache is wasted
and results in data being cached twice: first in the database cache and then in the
system buffer cache. By using direct I/O, Quick I/O does not waste memory on
double buffering. This frees up memory that can then be used by the database
server buffer cache, leading to increased performance.
About using Veritas File System files as raw character

devices
When VxFS with Quick I/O is installed, files may be accessed by the following
ways:
■ The VxFS interface treats the file as a regular VxFS file
■ The Quick I/O interface treats the same file as if it were a raw character device,
having performance similar to a raw device
Quick I/O allows a database server to use the Quick I/O interface while a backup
server uses the VxFS interface.
About the Quick I/O naming convention

To treat a file as a raw character device, Quick I/O requires a file name extension
to create an alias for a regular VxFS file. Quick I/O recognizes the alias when you
add the following suffix to a file name:
::cdev:vxfs:
The cdev portion is an acronym for character device. Whenever an application

opens an existing VxFS file with the suffix ::cdev:vxfs, Quick I/O treats the file as
if it were a raw device. For example, if the file xxx is a regular VxFS file, then an
application can access xxx as a raw character device by opening it with the name:
xxx::cdev:vxfs:
Note: When Quick I/O is enabled, you cannot create a regular VxFS file with a
name that uses the ::cdev:vxfs: extension. If an application tries to create a
regular file named xxx::cdev:vxfs:, the create fails. If Quick I/O is not available,
it is possible to create a regular file with the ::cdev:vxfs: extension, but this could
cause problems if Quick I/O is later enabled. Symantec advises you to reserve the
extension only for Quick I/O files.
About creating a Quick I/O file using qiomkfile
About use restrictions

There are restrictions to using regular VxFS files as Quick I/O files.
■ The name xxx::cdev:vxfs: is recognized as a special name by VxFS only
when the following conditions are met:
■ the qio module is loaded
■ Quick I/O has a valid license
■ the regular file xxx is physically present on the VxFS file system
■ there is no regular file named xxx::cdev:vxfs: on the system
■ If the file xxx is being used for memory mapped I/O, it cannot be accessed as
a Quick I/O file.
■ An I/O fails if the file xxx has a logical hole and the I/O is done to that hole on
xxx::cdev:vxfs:.
■ The size of the file cannot be extended by writes through the Quick I/O
interface.

The best way to make regular files accessible to the Quick I/O interface and
preallocate space for them is to use the qiomkfile command. Unlike the VxFS
setext command, which requires superuser privileges, any user who has read/write
permissions can run qiomkfile to create the files. The qiomkfile command has
five options:
-a Creates a symbolic link with an absolute path name for a specified file. The
default is to create a symbolic link with a relative path name.
-e (For Oracle database files to allow tablespace resizing.) Extends the file size
by the specified amount.
-h (For Oracle database files.) Creates a file with additional space allocated for
the Oracle header.
-r (For Oracle database files to allow tablespace resizing.) Increases the file to
the specified size.
-s Preallocates space for a file.
You can specify file size in terms of bytes (the default), or in kilobytes, megabytes,
gigabytes, or sectors (512 bytes) by adding a k, K, m, M, g, G, s, or S suffix. If the
size of the file including the header is not a multiple of the file system block size,
it is rounded to a multiple of the file system block size before preallocation.
The qiomkfile command creates two files: a regular file with preallocated,
contiguous space; and a symbolic link pointing to the Quick I/O name extension.
Creating a Quick I/O file using qiomkfile

The following example shows how to create a Quick I/O file using the qiomkfile
command.
See the qiomkfile(1) manual page.
To create a Quick I/O file using qiomkfile
1 Create a 100 MB file named dbfile in /database:
$ qiomkfile -s 100m /database/dbfile
The first file created is a regular file named /database/.dbfile, which has
the real space allocated. The second file is a symbolic link named
/database/dbfile. This is a relative link to /database/.dbfile via the Quick
I/O interface. That is, to .dbfile::cdev:vxfs:. This allows .dbfile to be
accessed by any database or application as a raw character device.
■ If you specify the -a option with qiomkfile, an absolute path name is
used, such as the following:
/database/dbfile points to /database/.dbfile::cdev:vxfs:
See “About absolute and relative path names” on page 190.

2 Check the results:
$ ls -al
-rw-r--r-- 1 oracle dba 104857600 Oct 22 15:03 .dbfile
lrwxrwxrwx 1 oracle dba 19 Oct 22 15:03 dbfile -> .dbfile::cdev:v
or:
$ ls -lL
crw-r----- 1 oracle dba 43,0 Oct 22 15:04 dbfile
■ If you specified the -a option with qiomkfile, the results are as follows:
$ ls -al
Accessing regular VxFS files through symbolic links
lrwxrwxrwx 1 oracle dba 31 Oct 22 15:05 dbfile ->

/database/.dbfile::cdev:vxfs:

One way to use Quick I/O is to create a symbolic link for each file in your database
and use the symbolic link to access the regular files as Quick I/O files. Any database
or application can then access the file as a raw character device.
See the Veritas Editions product documentation.
The following example creates a 100 MB Quick I/O file named dbfile on the VxFS
file system /database that can be accessed through a symbolic link.
To access a file through a symbolic link
1 Go to the /database file system:
$ cd /database
2 Create a 100 MB Quick I/O file named dbfile:
$ dd if=/dev/zero of=/database/.dbfile bs=128k count=800
The dd command preallocates the file space.

3 Create a symbolic link to dbfile:
$ ln -s .dbfile::cdev:vxfs: /database/dbfile
About absolute and relative path names

It is usually better to use relative path names instead of absolute path names when
creating symbolic links to access regular files as Quick I/O files. Using relative
path names prevents copies of the symbolic link from referring to the original
file. This is important if you are backing up or moving database files with a
command that preserves the symbolic link. However, some applications, such as
SAP, require absolute path names.
If you create a symbolic link using a relative path name, both the symbolic link
and the file are under the same parent directory. If you want to relocate the file,
both the file and the symbolic link must be moved.
It is also possible to use the absolute path name when creating a symbolic link. If
the database file is relocated to another directory, you must change the symbolic
link to use the new absolute path. You can put all the symbolic links in a directory
separate from the data directories. For example, you can create a directory named
/database and put in all the symbolic links, with the symbolic links pointing to
absolute path names.
Preallocating files using the setext command

You can use the VxFS setext command to preallocate file space, but the setext
command requires superuser privileges. You may need to use the chown and chgrp
commands to change the owner and group permissions on the file after it is
created.
See the setext(1) manual page.
The following example shows how to use setext to create a 100 MB database file
for an Oracle database.
To preallocate files using setext
1 Go to the /database file system:
# cd /database
2 Create the .dbfile file:
# touch .dbfile
3 Reserve 100 MB for the .dbfile file using setext:
# setext -r 102400 -f noreserve -f chgsize .dbfile
4 Create a symbolic link to .dbfile:
# ln -s .dbfile::cdev:vxfs: dbfile
5 Change the owner of dbfile to oracle:
# chown oracle dbfile
6 Change the group of dbfile to dba:
# chgrp dba dbfile

Using Quick I/O with Oracle databases
Using Quick I/O with Oracle databases

The following example shows how a file can be used by an Oracle database to
create a tablespace. This command would be run by the Oracle DBA, typically user
ID oracle. Oracle requires additional space for one Oracle header size. In the
following example, although 100 MB was allocated to /database/dbfile, the
Oracle database can use only up to 100 MB minus the Oracle parameter
db_block_size.
To create a tablespace with an Oracle database

1 Create the file dbfile and preallocate 100 MB for the file:
$ qiomkfile -h headersize -s 100m /database/dbfile
2 Start the Oracle database:
$ sqlplus /nolog
3 Create the tablespace:
SQL> connect / as sysdba

SQL> create tablespace ts1 datafile '/database/dbfile' size 99M;
SQL> exit;
Using Quick I/O with Sybase databases

To create a new database device, preallocate space on the file system by using the
qiomkfile command, then use the Sybase buildmaster command for a master
device, or the Transact SQL disk init command for a database device. qiomkfile
creates two files: a regular file using preallocated, contiguous space, and a symbolic
link pointing to the ::cdev:vxfs: name extension.
The following example creates a 100 megabyte master device masterdev on the
file system /sybmaster.
To create a new Sybase database device

1 Go to the /sybmaster file system:
$ cd /sybmaster
2 Create the masterdev file and preallocate 100 MB for the file:
$ qiomkfile -s 100m masterdev
You can use this master device while running the sybsetup program or
sybinit script.
3 To create the master device directly, enter:
$ buildmaster -d masterdev -s 51200
4 Add a new 500 megabyte database device datadev to the file system /sybdata
on your dataserver:
$ cd /sybdata
$ qiomkfile -s 500m datadev
...
5 Start the Sybase database:
$ isql -U sa -P sa_password -S dataserver_name
6 Set up the datadev database device:
1> disk init

2> name = "logical_name",
3> physname = "/sybdata/datadev",
4> vdevno = "device_number",
5> size = 256000
6> go

$ cd /sybmaster
You can use this master device while running the sybsetup program or sybinit
script.
on your dataserver:
$ cd /sybdata
...
1> disk init
5> size = 256000
6> go

$ cd /sybmaster
You can use this master device while running the sybsetup program or sybinit
script.
on your dataserver:
$ cd /sybdata
...
Enabling and disabling Quick I/O
1> disk init
5> size = 256000
6> go
Enabling and disabling Quick I/O

If the Quick I/O feature is licensed and installed, Quick I/O is enabled by default
when a file system is mounted. The -o qio and -o noqio mount options enable
and disable, respectively, Quick I/O when a file system is mounted.
If Quick I/O is not installed or licensed, a file system mounts by default without
Quick I/O and no error message is displayed. However, if you specify the -o qio
option, the mount command prints the following error message and terminates
without mounting the file system.
VxFDD: You don't have a license to run this program

vxfs mount: Quick I/O not available
To enable or disable Quick I/O

1 Specify the -o qio mount option to enable Quick I/O:
# mount -F vxfs -o qio MyFS
2 Specify the -o noqio mount option to disable Quick I/O:
# mount -F vxfs -o noqio MyFS

About Cached Quick I/O for databases

A 32-bit application (such as a 32-bit database) can use a maximum of only 4 GB
of memory because of the 32-bit address limitation. The Cached Quick I/O feature
improves database performance on machines with sufficient memory by also
using the file system cache to store data.
For read operations through the Quick I/O interface, data is cached in the system
page cache, so subsequent reads of the same data can access this cached copy and
avoid doing disk I/O. To maintain the correct data in its buffer for write operations,
Cached Quick I/O keeps the page cache in sync with the data written to disk.
With 64-bit applications, for which limited memory is not a critical problem, using
the file system cache still provides performance benefits by using the read-ahead
functionality. Because of the read-ahead functionality, sequential table scans will
benefit the most from using Cached Quick I/O by significantly reducing the query
response time.
Enabling Cached Quick I/O

Caching for Quick I/O files can be enabled online when the database is running
by using the vxtunefs utility and the qioadmin command.
See the vxtunefs(1M) and qioadmin(1) manual pages.
Note: Quick I/O must be enabled on the file system for Cached Quick I/O to operate.
To enable caching
1 Set the qio_cache_enable parameter of vxtunefs to enable caching on a file
system.
2 Enable the Cached Quick I/O feature for specific files using the qioadmin
command.
Enabling Cached Quick I/O for file systems

Caching is initially disabled on a file system. You enable Cached Quick I/O for a
file system by setting the qio_cache_enable option of the vxtunefs command
after the file system is mounted.
Note: The vxtunefs command enables caching for all the Quick I/O files on the
file system.
The following example enables Cached Quick I/O for the file system /database01.
To enable Cached Quick I/O for a file system

1 Enable Cached Quick I/O:
# vxtunefs -s -o qio_cache_enable=1 /database01
/database01 is a VxFS file system containing the Quick I/O files.
2 If desired, make this setting persistent across mounts by adding a file system
entry in the file /etc/vx/tunefstab:
/dev/vx/dsk/datadg/database01 qio_cache_enable=1
/dev/vx/dsk/datadg/database02 qio_cache_enable=1
See the tunefstab(4) manual page.
Manipulating Cached Quick I/O settings for individual files

A Quick I/O file's Cached Quick I/O settings are manipulated with the vxtunefs
utility and the qioadmin command.
See the vxtunefs(1M) and qioadmin(1) manual pages.
Note: The cache advisories operate only if Cached Quick I/O is enabled for the file
system. If the qio_cache_enable flag is zero, Cached Quick I/O is OFF for all the
files in that file system even if the individual file cache advisory for a file is ON.
To enable caching on a file

◆ Enable caching on a file:
$ qioadmin -S filename=on mount_point
To disable caching on a file

◆ Disable caching on a file:
$ qioadmin -S filename=off mount_point

About Quick I/O statistics
To make the caching setting persistent across mounts

◆ Create a qiotab file, /etc/vx/qioadmin, to list files and their caching
advisories. Based on the following example, the file /database/sell.dbf will
have caching turned on whenever the file system /database is mounted:
device=/dev/vx/dsk/datadg/database01
dates.dbf,off
names.dbf,off
sell.dbf,on
To check on the current cache advisory settings for a file

◆ Check the current cache advisory settings:
$ qioadmin -P filename mount_point
filename,OFF
To check the setting of the qio_cache_enable flag for a file system

◆ Check the setting of the qio_cache_enable flag:
$ vxtunefs -p /database01
qio_cache_enable = 1
Check the setting of the flag qio_cache_enable using the vxtunefs command,
and the individual cache advisories for each file, to verify caching.
About Quick I/O statistics

Quick I/O provides the qiostat utility to collect database I/O statistics generated
over a period of time. qiostat reports statistics, such as the number of read and
write operations, the number of blocks read or written, and the average time spent
on read and write operations during an interval.
See the qiostat(1) manual page.
Increasing database performance using Quick I/O

Perform the following steps to increase database performance on a VxFS file
system using Quick I/O.
See the Veritas Editions product documentation.
See the qioadmin(1) and vxtunefs(1M) manual pages.
Increasing database performance using Quick I/O
To increase database performance

1 Verify that the Quick I/O module is loaded.
# modinfo | grep fdd
2 You can add the following line to the file /etc/system to load Quick I/O
whenever the system reboots.
forceload: drv/fdd
3 Create a regular VxFS file and preallocate it to the required size, or use the
qiomkfile command. The size of this preallocation depends on the size
requirement of the database server.
4 Create and access the database using the file name xxx::cdev:vxfs:.
Appendix A
Quick Reference
This appendix includes the following topics:
■ Command summary
■ Online manual pages
■ Creating a VxFS file system
■ Converting a file system to VxFS
■ Mounting a file system
■ Unmounting a file system
■ Displaying information on mounted file systems
■ Identifying file system types
■ Identifying file system types
■ Resizing a file system
■ Backing up and restoring a file system
■ Using quotas
Command summary
Symbolic links to all VxFS command executables are installed in the /opt/VRTS/bin
directory. Add this directory to the end of your PATH environment variable to
access the commands.
Table A-1 describes the VxFS-specific commands.
202 Quick Reference
Command summary
Table A-1 VxFS commands
Command Description
cfscluster CFS cluster configuration command. This functionality is available only with the Veritas Cluster
File System product.
cfsdgadm Adds or deletes shared disk groups to or from a cluster configuration. This functionality is
available only with the Veritas Cluster File System product.
cfsmntadm Adds, deletes, modifies, and sets policy on cluster mounted file systems. This functionality is
cfsmount, Mounts or unmounts a cluster file system. This functionality is available only with the Veritas
cfsumount Cluster File System product.
cp Copies files and directories on VxFS file systems.
df Reports the number of free disk blocks and inodes for a VxFS file system.
fcladm Administers VxFS File Change Logs.
ff Lists file names and inode information for a VxFS file system.
fiostat Administers file I/O statistics
fsadm Resizes or defragments a VxFS file system.
fsapadm Administers VxFS allocation policies.
fscat Cats a VxFS file system.
fscdsadm Performs online CDS operations.
fscdsconv Performs offline CDS migration tasks on VxFS file systems.
fscdstask Performs various CDS operations.
fsck Checks and repairs a VxFS file system.
fsckpt_restore Restores file systems from VxFS Storage Checkpoints.
fsckptadm Administers VxFS Storage Checkpoints.
fsclustadm Manages cluster-mounted VxFS file systems. This functionality is available only with the Veritas
Cluster File System product.
fsdb Debugs VxFS file systems.
fsdbencap Encapsulates databases.
fsmap Displays VxFS file system extent information.

Quick Reference 203
Command summary
Table A-1 VxFS commands (continued)
Command Description
fsppadm Administers VxFS placement policies.
fstyp Returns the type of file system on a specified disk partition.
fsvmap Maps volumes of VxFS file systems to files.
fsvoladm Administers VxFS volumes.
glmconfig Configures Group Lock Managers (GLM).
This functionality is available only with the Veritas Cluster File System product.
ls Lists files and directories on a VxFS file system.
mkfs Constructs a VxFS file system.
mount Mounts a VxFS file system.
mv Moves files and directories on a VxFS file system.
ncheck Generates path names from inode numbers for a VxFS file system.
qioadmin Administers VxFS Quick I/O for Databases cache.
qiomkfile Creates a VxFS Quick I/O device file. This functionality is available only with the Veritas Quick
I/O for Databases feature.
qiostat Displays statistics for VxFS Quick I/O for Databases. This functionality is available only with
the Veritas Quick I/O for Databases feature.
qlogadm Administers low level IOCTL for the QuickLog driver. This functionality is available only with
the Veritas QuickLog feature.
qlogattach Attaches a previously formatted QuickLog volume to a QuickLog device. This functionality is
available only with the Veritas QuickLog feature.
qlogck Recovers QuickLog devices during the boot process. This functionality is available only with
qlogclustadm Administers Cluster QuickLog devices. This functionality is available only with the Veritas
QuickLog feature.
qlogdb Debugs QuickLog. This functionality is available only with the Veritas QuickLog feature.
qlogdetach Detaches a QuickLog volume from a QuickLog device. This functionality is available only with
204 Quick Reference
Command summary
Command Description
qlogdisable Remounts a VxFS file system with QuickLog logging disabled. This functionality is available
only with the Veritas QuickLog feature.
qlogenable Remounts a VxFS file system with QuickLog logging enabled. This functionality is available
qlogmk Creates and attaches a QuickLog volume to a QuickLog device. This functionality is available
qlogprint Displays records from the QuickLog configuration. This functionality is available only with the
Veritas QuickLog feature.
qlogrec Recovers the QuickLog configuration file during a system failover. This functionality is available
qlogrm Removes a QuickLog volume from the configuration file. This functionality is available only
with the Veritas QuickLog feature.
qlogstat Prints statistics for running QuickLog devices, QuickLog volumes, and VxFS file systems. This
functionality is available only with the Veritas QuickLog feature.
qlogtrace Prints QuickLog tracing. This functionality is available only with the Veritas QuickLog feature.
setext Sets extent attributes on a file in a VxFS file system.
umount_vxfs Unmounts a VxFS file system.
vxdump Incrementally dumps file systems.
vxedquota Edits user quotas for a VxFS file system.
vxenablef Enables specific VxFS features.
vxfsconvert Converts an unmounted file system to VxFS or upgrades a VxFS disk layout version.
vxfsstat Displays file system statistics.
vxlsino Looks up VxFS reverse path names.
vxquot Displays file system ownership summaries for a VxFS file system.
vxquota Displays user disk quotas and usage on a VxFS file system.
vxquotaoff Turns quotas on and off for a VxFS file system.

vxquotaon
vxrepquota Summarizes quotas for a VxFS file system.

Quick Reference 205
Online manual pages
Command Description
vxrestore Restores a file system incrementally.
vxtunefs Tunes a VxFS file system.
vxupgrade Upgrades the disk layout of a mounted VxFS file system.
Online manual pages

This release includes the following online manual pages as part of the VRTSvxfs
package. These are installed in the appropriate directories under /opt/VRTS/man
(add this to your MANPATH environment variable), but does not update the windex
database. To ensure that new VxFS manual pages display correctly, update the
windex database after installing VRTSvxfs.
See the catman(1M) manual page.
Table A-2 describes the VxFS-specific section 1 manual pages.
Table A-2 Section 1 manual pages
Section 1 Description
cp_vxfs Copies files and directories on a VxFS file system.
cpio_vxfs Copies files and directories on a VxFS file system.
fiostat Administers file I/O statistics.
getext Gets extent attributes for a VxFS file system.
ls_vxfs Lists files and directories on a VxFS file system.
mv_vxfs Moves files and directories on a VxFS file system.
qioadmin Administers VxFS Quick I/O for Databases cache. This functionality is available only with the
Veritas Quick I/O for Databases feature.
qiomkfile Creates a VxFS Quick I/O device file. This functionality is available only with the Veritas Quick
I/O for Databases feature.
qiostat Displays statistics for VxFS Quick I/O for Databases. This functionality is available only with
the Veritas Quick I/O for Databases feature.
setext Sets extent attributes on a file in a VxFS file system.

206 Quick Reference
Online manual pages
Table A-3 describes the VxFS-specific section 1M manual pages.
Table A-3 Section 1M manual pages
Section 1M Description
cfscluster Configures CFS clusters. This functionality is available only with the Veritas Cluster File System
product.
cfsdgadm Adds or deletes shared disk groups to/from a cluster configuration. This functionality is available
only with the Veritas Cluster File System product.
cfsmntadm Adds, deletes, modifies, and sets policy on cluster mounted file systems. This functionality is
cfsmount, Mounts or unmounts a cluster file system. This functionality is available only with the Veritas
cfsumount Cluster File System product.
df_vxfs Reports the number of free disk blocks and inodes for a VxFS file system.
fcladm Administers VxFS File Change Logs.
ff_vxfs Lists file names and inode information for a VxFS file system.
fsadm_vxfs Resizes or reorganizes a VxFS file system.
fsapadm Administers VxFS allocation policies.
fscat_vxfs Cats a VxFS file system.
fscdsadm Performs online CDS operations.
fscdsconv Performs offline CDS migration tasks on VxFS file systems.
fscdstask Performs various CDS operations.
fsck_vxfs Checks and repairs a VxFS file system.
fsckptadm Administers VxFS Storage Checkpoints.
fsckpt_restore Restores file systems from VxFS Storage Checkpoints.
fsclustadm Manages cluster-mounted VxFS file systems. This functionality is available only with the Veritas
fsdbencap Encapsulates databases.
fsdb_vxfs Debugs VxFS file systems.
fsmap Displays VxFS file system extent information.
fsppadm Administers VxFS placement policies.

Quick Reference 207
Online manual pages
Table A-3 Section 1M manual pages (continued)
fstyp_vxfs Returns the type of file system on a specified disk partition.
fsvmap Maps volumes of VxFS file systems to files.
fsvoladm Administers VxFS volumes.
glmconfig Configures Group Lock Managers (GLM). This functionality is available only with the Veritas
mkfs_vxfs Constructs a VxFS file system.
mount_vxfs Mounts a VxFS file system.
ncheck_vxfs Generates path names from inode numbers for a VxFS file system.
qlogadm Administers low level IOCTL for the QuickLog driver. This functionality is available only with
qlogck Recovers QuickLog devices during the boot process. This functionality is available only with
qlogdetach Detaches a QuickLog volume from a QuickLog device. This functionality is available only with
qlogmk Creates and attaches a QuickLog volume to a QuickLog device. This functionality is available
qlogstat Prints statistics for running QuickLog devices, QuickLog volumes, and VxFS file systems. This
functionality is available only with the Veritas QuickLog feature.
quot Summarizes ownership on a VxFS file system.
quotacheck_vxfs Checks VxFS file system quota consistency.
umount_vxfs Unmounts a VxFS file system.
vxdiskusg Generates VxFS disk accounting data by user ID.
vxdump Incrementally dumps file systems.
vxedquota Edits user quotas for a VxFS file system.
vxenablef Enables specific VxFS features.
vxfsconvert Converts an unmounted file system to VxFS or upgrades a VxFS disk layout version.
vxfsstat Displays file system statistics.

208 Quick Reference
Online manual pages
Table A-3 Section 1M manual pages (continued)
vxlsino Looks up VxFS reverse path names.
vxquot Displays file system ownership summaries for a VxFS file system.
vxquota Displays user disk quotas and usage on a VxFS file system.
vxquotaoff Turns quotas on and off for a VxFS file system.

vxquotaon
vxrepquota Summarizes quotas for a VxFS file system.
vxrestore Restores a file system incrementally.
vxtunefs Tunes a VxFS file system.
vxupgrade Upgrades the disk layout of a mounted VxFS file system.
vxfs_ap_alloc2 Allocates an fsap_info2 structure.
vxfs_ap_assign_ckpt Assigns an allocation policy to file data and metadata in a Storage

Checkpoint.
vxfs_ap_assign_file Assigns an allocation policy for file data and metadata.
vxfs_ap_assign_file_pat Assigns a pattern-based allocation policy for a directory.
vxfs_ap_assign_fs Assigns an allocation policy for all file data and metadata within a
specified file system.
vxfs_ap_assign_fs_pat Assigns an pattern-based allocation policy for a file system.
vxfs_ap_define Defines a new allocation policy.
vxfs_ap_define2 Defines a new allocation policy.
vxfs_ap_enforce_file Ensures that all blocks in a specified file match the file allocation policy.
vxfs_ap_enforce_file2 Reallocates blocks in a file to match allocation policies.
vxfs_ap_enumerate Returns information about all allocation policies.
vxfs_ap_enumerate2 Returns information about all allocation policies.

Quick Reference 209
Online manual pages
Table A-4 Section 3 manual pages (continued)
vxf_ap_free2 Frees one or more fsap_info2 structures.
vxfs_ap_query Returns information about a specific allocation policy.
vxfs_ap_query2 Returns information about a specific allocation policy.
vxfs_ap_query_ckpt Returns information about allocation policies for each Storage

Checkpoint.
vxfs_ap_query_file Returns information about allocation policies assigned to a specified

file.
vxfs_ap_query_file_pat Returns information about the pattern-based allocation policy assigned

to a directory.
vxfs_ap_query_fs Retrieves allocation policies assigned to a specified file system.
vxfs_ap_query_fs_pat Returns information about the pattern-based allocation policy assigned

to a file system.
vxfs_ap_remove Deletes a specified allocation policy.
vxfs_fcl_sync Sets a synchronization point in the VxFS File Change Log.
vxfs_fiostats_dump Returns file and sub-file I/O statistics.
vxfs_fiostats_getconfig Gets sub-file I/O statistics configuration values.
vxfs_fiostats_set Turns on and off sub-file I/O statistics and resets statistics counters.
vxfs_get_ioffsets Obtains VxFS inode field offsets.
vxfs_inotopath Returns path names for a given inode number.
vxfs_nattr_check Checks for the existence of named data streams.
vxfs_nattr_fcheck
vxfs_nattr_link Links to a named data stream.
vxfs_nattr_open Opens a named data stream.
vxfs_nattr_rename Renames a named data stream.
vxfs_nattr_unlink Removes a named data stream.
vxfs_nattr_utimes Sets access and modification times for named data streams.
vxfs_vol_add Adds a volume to a multi-volume file system.

210 Quick Reference
Online manual pages
Table A-4 Section 3 manual pages (continued)
vxfs_vol_clearflags Clears specified flags on volumes in a multi-volume file system.
vxfs_vol_deencapsulate De-encapsulates a volume from a multi-volume file system.
vxfs_vol_encapsulate Encapsulates a volume within a multi-volume file system.
vxfs_vol_encapsulate_bias Encapsulates a volume within a multi-volume file system.
vxfs_vol_enumerate Returns information about the volumes within a multi-volume file

system.
vxfs_vol_queryflags Queries flags on volumes in a multi-volume file system.
vxfs_vol_remove Removes a volume from a multi-volume file system.
vxfs_vol_resize Resizes a specific volume within a multi-volume file system.
vxfs_vol_setflags Sets specified flags on volumes in a multi-volume file system.
vxfs_vol_stat Returns free space information about a component volume within a

multi-volume file system.
fs_vxfs Provides the format of a VxFS file system volume.
inode_vxfs Provides the format of a VxFS file system inode.
qlog_config Describes the QuickLog configuration file. This functionality is available only with the Veritas
QuickLog feature.
tunefstab Describes the VxFS file system tuning parameters table.
qlog Describes the Veritas QuickLog device driver. This functionality is available only with the
Veritas QuickLog feature.
vxfsio Describes the VxFS file system control functions.

Quick Reference 211
Creating a VxFS file system
Creating a VxFS file system

The mkfs command creates a VxFS file system by writing to a special character
device file. The special character device is a Veritas Volume Manager (VxVM)
volume. mkfs builds a file system with a root directory and a lost+found directory.
Before running mkfs, you must create the target device. Refer to your operating
system documentation for more information. If you are using a logical device
(such as a VxVM volume), see the VxVM documentation for instructions on device
initialization.
To create a file system
◆ mkfs [-F vxfs] [generic_options] [-o specific_options] \
special [size]
vxfs The file system type.
generic_options Options common to most other file system types.
specific_options Options specific to VxFS.
-o N Displays the geometry of the file system and does not write
to the device.
-o largefiles Allows users to create files larger than two gigabytes. The
default option is largefiles.
special The Veritas Volume Manager volume.The Veritas Volume

Manager volume.The Veritas Volume Manager volume. The
block device or Veritas Volume Manager volume.The Veritas
Volume Manager volume.
size The size of the new file system (in sectors).
Example of creating a file system

The following example creates a VxFS file system of 12288 sectors in size on a
VxVM volume.
212 Quick Reference
Converting a file system to VxFS
To create a VxFS file system

1 Create the file system:
# mkfs -F vxfs /dev/vx/rdsk/diskgroup/volume 12288

version 7 layout
12288 sectors, 6144 blocks of size 1024, log size 512 blocks
2 Mount the newly created file system.
Converting a file system to VxFS

The vxfsconvert command can be used to convert a UFS file system to a VxFS
file system.
See the vxfsconvert(1M) manual page.
To convert a UFS file system to a VxFS file system
◆ Use the vxfsconvert command to convert a UFS file system to VxFS:
vxfsconvert [-l logsize] [-s size] [-efnNvyY] special
-e Estimates the amount of space required to complete the conversion.
-f Displays the list of supported file system types.
-l logsize Specifies the size of the file system intent log.
-n|N Assumes a no response to all questions asked by vxfsconvert.
-s size Directs vxfsconvert to use free disk space past the current end of the
file system to store VxFS metadata.
-v Specifies verbose mode.
-y|Y Assumes a yes response to all questions asked by vxfsconvert.
special Specifies the name of the character (raw) device that contains the file
system to convert.
Example of converting a file system

The following example converts a UFS file system to a VxFS file system with an
intent log size of 4096 blocks.
Quick Reference 213
Mounting a file system
To convert a UFS file system to a VxFS file system

◆ Convert the file system:
# vxfsconvert -l 4096 /dev/vx/rdsk/diskgroup/volume

You can mount a VxFS file system by using the mount command. When you enter
the mount command, the generic mount command parses the arguments and the
-F FSType option executes the mount command specific to that file system type.
The mount command first searches the /etc/fs/FSType directory, then the
/usr/lib/fs/FSType directory. If the -F option is not supplied, the command
searches the file /etc/vfstab for a file system and an FSType matching the special
file or mount point provided. If no file system type is specified, mount uses the
default file system.
To mount a file system
◆ Use the mount command to mount a file system:
mount [-F vxfs] [generic_options] [-r] [-o specific_options] \

special mount_point
vxfs File system type.
generic_options Options common to most other file system types.
specific_options Options specific to VxFS.
-o ckpt=ckpt_name Mounts a Storage Checkpoint.
-o cluster Mounts a file system in shared mode. Available only with the VxFS
cluster file system feature.
-o cluster Mounts a file system in shared mode. Available only with the VxFS
cluster file system feature.
special Block special device.
mount_point Directory on which to mount the file system.
-r Mounts the file system as read-only.

214 Quick Reference
Mount options
The mount command has numerous options to tailor a file system for various
functions and environments.
The following table lists some of the specific_options:
Security feature If security is important, use blkclear to ensure that deleted files
are completely erased before the space is reused.
Support for large files If you specify the largefiles option, you can create files larger than
two gigabytes on the file system. The default option is largefiles.
Support for cluster file If you specify the cluster option, the file system is mounted in
systems shared mode. Cluster file systems depend on several other Veritas
products that must be correctly configured before a complete
clustering environment is enabled.
Using Storage The ckpt=checkpoint_name option mounts a Storage

Checkpoints Checkpoint of a mounted file system that was previously created
by the fsckptadm command.
Using databases If you are using databases with VxFS and if you have installed a
license key for the Veritas Quick I/O for Databases feature, the
mount command enables Quick I/O by default (the same as
specifying the qio option). The noqio option disables Quick I/O. If
you do not have Quick I/O, mount ignores the qio option.
Alternatively, you can increase database performance using the
mount option convosync=direct, which utilizes direct I/O.
See “About Quick I/O” on page 185.
News file systems If you are using cnews, use delaylog (or
tmplog),mincache=closesync because cnews does an fsync() on
each news file before marking it received. The fsync() is
performed synchronously as required, but other options are
delayed.
Temporary file For a temporary file system such as /tmp, where performance is
systems more important than data integrity, use
tmplog,mincache=tmpcache.
See “Choosing mount command options” on page 30.

See the fsckptadm(1M), mount(1M), mount_vxfs(1M), and vfstab(4) manual pages.
Quick Reference 215
Example of mounting a file system

The following example mounts the file system /dev/vx/dsk/fsvol/vol1 on the
/ext directory with read/write access and delayed logging.
To mount the file system

◆ Mount the file system:
# mount -F vxfs -o delaylog /dev/vx/dsk/fsvol/vol1 /ext
Editing the vfstab file

You can edit the /etc/vfstab file to mount a file system automatically at boot
time.
You must specify the following:
■ The special block device name to mount
■ The special character device name used by fsck
■ The mount point
■ The mount options
■ The file system type (vxfs)
■ Which fsck pass looks at the file system
■ Whether to mount the file system at boot time
Each entry must be on a single line.
See the vfstab(4) manual page.
The following is a typical vfstab file with the new file system on the last line:
# device # to mount # device to fsck mount point FS type fsck mount at mount
pass boot options
# /dev/dsk/c1d0s2 /dev/rdsk/c1d0s2 /usr ufs 1 yes —
/proc — /proc proc — no —
fd — /dev/fd fd — no —
swap — /tmp tmpfs — yes —
/dev/dsk/c0t3d0s0 /dev/rdsk/c0t3d0s0 / ufs 1 no —
/dev/dsk/c0t3d0s1 — — swap — no —
216 Quick Reference
Unmounting a file system
/dev/vx/dsk/fsvol/vol1 /dev/vx/rdsk/fsvol/vol1 /ext vxfs 1 yes —
Unmounting a file system

Use the umount command to unmount a currently mounted file system.
To unmount a file system
◆ Use the umount command to unmount a file system:
umount special mount_point
Specify the file system to be unmounted as a mount_point or special. special

is the device on which the file system resides.
Example of unmounting a file system

The following are examples of unmounting file systems.
To unmount the file system /dev/vx/dsk/fsvol/vol1
◆ Unmount the file system:
# umount /dev/vx/dsk/fsvol/vol1
To unmount all file systems not required by the system

◆ Unmount the file systems:
# umount -a
This unmounts all file systems except /, /usr, /usr/kvm, /var, /proc, /dev/fd,
and /tmp.
Displaying information on mounted file systems

Use the mount command to display a list of currently mounted file systems.
See the mount(1M) and mount_vxfs(1M) manual pages.
Quick Reference 217
Identifying file system types
To view the status of mounted file systems

◆ Use the mount command to view the status of mounted file systems:
mount -v
This shows the file system type and mount options for all mounted file systems.
The -v option specifies verbose mode.
To view the status of mounted file systems
◆ Use the mount command to view the status of mounted file systems:
mount -v
This shows the file system type and mount options for all mounted file systems.
The -v option specifies verbose mode.
Example of displaying information on mounted file systems

The following example shows the result of invoking the mount command without
options.
To display information on mounted file systems
◆ Invoke the mount command without options:
# mount
/ on /dev/root read/write/setuid on Thu May 26 16:58:24 2004
/proc on /proc read/write on Thu May 26 16:58:25 2004
/dev/fd on /dev/fd read/write on Thu May 26 16:58:26 2004
/tmp on /tmp read/write on Thu May 26 16:59:33 2004
/var/tmp on /var/tmp read/write on Thu May 26 16:59:34 2004

Use the fstyp command to determine the file system type for a specified file
system. This is useful when a file system was created elsewhere and you want to
know its type.
See the fstyp(1M) and fstyp_vxfs(1M) manual pages.
To determine a file system's type
◆ Use the fstyp command to determine a file system's type:
fstyp -v special
218 Quick Reference
special The character (raw) device.
Example of determining a file system's type

The following example uses the fstyp command to determine a the file system
type of the /dev/vx/dsk/fsvol/vol1 device.
To determine the file system's type
◆ Use the fstyp command to determine the file system type of the device
/dev/vx/dsk/fsvol/vol1:
# fstyp -v /dev/vx/dsk/fsvol/vol1
The output indicates that the file system type is vxfs, and displays file system
information similar to the following:
vxfs
magic a501fcf5 version 7 ctime Tue Jun 25 18:29:39 2003
logstart 17 logend 1040
bsize 1024 size 1048576 dsize 1047255 ninode 0 nau 8
defiextsize 64 ilbsize 0 immedlen 96 ndaddr 10
aufirst 1049 emap 2 imap 0 iextop 0 istart 0
bstart 34 femap 1051 fimap 0 fiextop 0 fistart 0 fbstart
1083
nindir 2048 aulen 131106 auimlen 0 auemlen 32
auilen 0 aupad 0 aublocks 131072 maxtier 17
inopb 4 inopau 0 ndiripau 0 iaddrlen 8 bshift 10
inoshift 2 bmask fffffc00 boffmask 3ff checksum d7938aa1
oltext1 9 oltext2 1041 oltsize 8 checksum2 52a
free 382614 ifree 0
efree 676 413 426 466 612 462 226 112 85 35 14 3 6 5 4 4 0 0

Use the fstyp command to determine the file system type for a specified file
system. This is useful when a file system was created elsewhere and you want to
know its type.
See the fstyp(1M) and fstyp_vxfs(1M) manual pages.
Quick Reference 219
To determine a file system's type

◆ Use the fstyp command to determine a file system's type:
fstyp -v special
special The character (raw) device.
Example of determining a file system's type

The following example uses the fstyp command to determine a the file system
type of the /dev/vx/dsk/fsvol/vol1 device.
To determine the file system's type
◆ Use the fstype command to determine the file system type of the device
/dev/vx/dsk/fsvol/vol1:
# fstyp -v /dev/vx/dsk/fsvol/vol1
The output indicates that the file system type is vxfs, and displays file system
information similar to the following:
vxfs
magic a501fcf5 version 6 ctime Tue Jun 25 18:29:39 2003
logstart 17 logend 1040
bsize 1024 size 1048576 dsize 1047255 ninode 0 nau 8
defiextsize 64 ilbsize 0 immedlen 96 ndaddr 10
aufirst 1049 emap 2 imap 0 iextop 0 istart 0
bstart 34 femap 1051 fimap 0 fiextop 0 fistart 0 fbstart
1083
nindir 2048 aulen 131106 auimlen 0 auemlen 32
auilen 0 aupad 0 aublocks 131072 maxtier 17
inopb 4 inopau 0 ndiripau 0 iaddrlen 8 bshift 10
inoshift 2 bmask fffffc00 boffmask 3ff checksum d7938aa1
oltext1 9 oltext2 1041 oltsize 8 checksum2 52a
free 382614 ifree 0
efree 676 413 426 466 612 462 226 112 85 35 14 3 6 5 4 4 0 0
220 Quick Reference
Resizing a file system

You can extend or shrink mounted VxFS file systems using the fsadm command.
Use the extendfs command to extend the size of an unmounted file system. A
file system using the Version 4 disk layout can be up to two terabytes in size. A
file system using the Version 5 disk layout can be up to 32 terabytes in size. A file
system using the Version 6 or 7 disk layout can be up to 8 exabytes in size. The
size to which a Version 5, 6, or 7 disk layout file system can be increased depends
on the file system block size.
See “About disk layouts” on page 277.
See the format(1M) and fsadm_vxfs(1M) manual pages.
Extending a file system using fsadm

If a VxFS file system is not large enough, you can increase its size. The size of the
file system is specified in units of 512-byte blocks (or sectors).
Note: If a file system is full, busy, or too fragmented, the resize operation may
fail.
The device must have enough space to contain the larger file system.
See the format(1M) manual page.
See the Veritas Volume Manager Administrator's Guide.
To extend a VxFS file system
◆ Use the fsadm command to extend a VxFS file system:
/usr/lib/fs/vxfs/fsadm [-b newsize] [-r rawdev] \

mount_point
newsize The size (in sectors) to which the file system will increase.
mount_point The file system's mount point.
-r rawdev Specifies the path name of the raw device if there is no entry in
/etc/vfstab and fsadm cannot determine the raw device.
Example of extending a file system

The following is an example of extending a file system with the fsadm command.
Quick Reference 221
To extend a file system

◆ Extend the VxFS file system mounted on /ext to 22528 sectors:
# fsadm -b 22528 /ext
Shrinking a file system

You can decrease the size of the file system using fsadm, even while the file system
is mounted.
Note: In cases where data is allocated toward the end of the file system, shrinking
may not be possible. If a file system is full, busy, or too fragmented, the resize
operation may fail.
To decrease the size of a VxFS file system

◆ Use the fsadm command to decrease the size of a VxFS file system:
fsadm [-b newsize] [-r rawdev] mount_point
newsize The size (in sectors) to which the file system will shrink.
Specifies the path name of the raw device if there is no entry in

/etc/fstab and fsadm cannot determine the raw device.

/etc/filesystems and fsadm cannot determine the raw device.
Example of shrinking a file system

The following example shrinks a VxFS file system mounted at /ext to 20480
sectors.
222 Quick Reference
To shrink a VxFS file system

◆ Shrink a VxFS file system mounted at /ext to 20480 sectors:
# fsadm -b 20480 /ext
Warning: After this operation, there is unused space at the end of the device.
You can then resize the device, but be careful not to make the device smaller
than the new size of the file system.
Reorganizing a file system

You can reorganize (or compact) a fragmented file system using fsadm, even while
the file system is mounted. This may help shrink a file system that could not
previously be decreased.
Note: If a file system is full or busy, the reorg operation may fail.
To reorganize a VxFS file system

◆ Use the fsadm command to reorganize a VxFS file system:
fsadm [-e] [-d] [-E] [-D] [-r rawdev] mount_point
-d Reorders directory entries to put subdirectory entries first, then

all other entries in decreasing order of time of last access. Also
compacts directories to remove free space.
-D Reports on directory fragmentation.
-e Minimizes file system fragmentation. Files are reorganized to

have the minimum number of extents.
-E Reports on extent fragmentation.

/etc/fstab and fsadm cannot determine the raw device.

/etc/filesystems and fsadm cannot determine the raw device.
Quick Reference 223
Backing up and restoring a file system
Example of reorganizing a file system

The following example reorganizes the file system mounted at /ext.
To reorganize a VxFS file system
◆ Reorganize the VxFS file system mounted at /ext:
# fsadm -EeDd /ext

To back up a VxFS file system, you first create a read-only snapshot file system,
then back up the snapshot. This procedure lets you keep the main file system on
line. The snapshot is a copy of the snapped file system that is frozen at the moment
the snapshot is created.
See “About snapshot file systems” on page 93.
See the mount(1M), mount_vxfs(1M), vxdump(1M), and vxrestore(1M) manual
pages.
Creating and mounting a snapshot file system

The first step in backing up a VxFS file system is to create and mount a snapshot
file system.
To create and mount a snapshot of a VxFS file system
◆ Use the mount command to create and mount a snapshot of a VxFS file system:
mount [-F vxfs] -o snapof=source,[snapsize=size] \

destination snap_mount_point
source The special device name or mount point of the file system to
copy.
destination The name of the special device on which to create the snapshot.
size The size of the snapshot file system in sectors.
snap_mount_point Location where to mount the snapshot; snap_mount_pointmust

exist before you enter this command.
224 Quick Reference
Example of creating and mounting a snapshot of a VxFS file

system
The following example creates a snapshot file system of the file system at /home
on /dev/vx/dsk/fsvol/vol1, and mounts it at /snapmount.
To create and mount a snapshot file system of a file system
◆ Create a snapshot file system of the file system at /home on
/dev/vx/dsk/fsvol/vol1 and mount it at /snapmount:
# mount -F vxfs -o snapof=/dev/vx/dsk/fsvol/vol1, \

snapsize=32768 /dev/vx/dsk/fsvol/vol1 /snapmount
You can now back up the file system.
Backing up a file system

After creating a snapshot file system, you can use vxdump to back it up.
To back up a VxFS snapshot file system
◆ Use the vxdump command to back up a VxFS snapshot file system:
vxdump [-c] [-f backupdev] snap_mount_point
-c Specifies using a cartridge tape device.
backupdev The device on which to back up the file system.
snap_mount_point The snapshot file system's mount point.
Example of backing up a file system

The following example backs up the VxFS snapshot file system mounted at
/snapmount to the tape drive with device name /dev/rmt/00m.
To back up a VxFS snapshot file system

◆ Back up the VxFS snapshot file system mounted at /snapmount to the tape
drive with device name /dev/rmt/00m:
# vxdump -cf /dev/rmt/00m /snapmount

Quick Reference 225
Using quotas
Restoring a file system

After backing up the file system, you can restore it using the vxrestore command.
First, create and mount an empty file system.
To restore a VxFS snapshot file system
◆ Use the vxrestore command to restore a VxFS snapshot file system:
vxrestore [-v] [-x] [filename]
-x Extracts the named files from the tape.
filename The file or directory to restore. If filename is omitted, the root

directory, and thus the entire tape, is extracted.
Example of restoring a file system

The following example restores a VxFS snapshot file system from the tape:
/dev/st1 into the mount point /restore
To restore a VxFS snapshot file system

◆ Restore a VxFS snapshot file system from the tape /dev/st1 into the mount
point /restore:
# cd /restore
# vxrestore -v -x -f /dev/st1
Using quotas
You can use quotas to allocate per-user quotas on VxFS file systems.
See “Using quotas” on page 104.
See the vxquota(1M), vxquotaon(1M), vxquotaoff(1M), and vxedquota(1M) manual
pages.
Turning on quotas
You can enable quotas at mount time or after a file system is mounted. The root
directory of the file system must contain a file named quotas that is owned by
root.
226 Quick Reference
Using quotas
To turn on quotas
1 Turn on quotas for a mounted file system:
vxquotaon mount_point
2 Mount a file system and turn on quotas at the same time:
mount -F vxfs -o quota special mount_point
If the root directory does not contain a quotas file, the mount command
succeeds, but quotas are not turned on.
Example of turning on quotas for a mounted file system

The following example creates a quoatas file and turns on quotas for a VxFS file
system mounted at /mnt.
To turn on quotas for a mounted file system
◆ Create a quotas file if it does not already exist and turn on quotas for a VxFS
file system mounted at /mnt:
# touch /mnt/quotas
# vxquotaon /mnt
Example of turning on quotas at mount time

The following example turns on quotas when the /dev/vx/dsk/fsvol/vol1 file
system is mounted.
To turn on quotas for a file system at mount time
◆ Turn on quotas at mount time by specifying the -o quota option:
# mount -F vxfs -o quota /dev/vx/dsk/fsvol/vol1 /mnt
Setting up user quotas

You can set user quotas with the vxedquota command if you have superuser
privileges. User quotas can have a soft limit and hard limit. You can modify the
limits or assign them specific values. Users are allowed to exceed the soft limit,
but only for a specified time. Disk usage can never exceed the hard limit. The
default time limit for exceeding the soft limit is seven days on VxFS file systems.
Quick Reference 227
Using quotas
vxedquota creates a temporary file for a specified user. This file contains on-disk
quotas for each mounted VxFS file system that has a quotas file. The temporary
file has one or more lines similar to the following:
fs /mnt blocks (soft = 0, hard = 0) inodes (soft=0, hard=0)

fs /mnt1 blocks (soft = 100, hard = 200) inodes (soft=10, hard=20)
Quotas do not need to be turned on for vxedquota to work. However, the quota
limits apply only after quotas are turned on for a given file system.
vxedquota has an option to modify time limits. Modified time limits apply to the
entire file system; you cannot set time limits for an individual user.
To set up user quotas
1 Invoke the quota editor:
vxedquota username
2 Modify the time limit:
vxedquota -t
Viewing quotas
The superuser or individual user can view disk quotas and usage on VxFS file
systems using the vxquota command. This command displays the user's quotas
and disk usage on all mounted VxFS file systems where the quotas file exists. You
will see all established quotas regardless of whether or not the quotas are actually
turned on.
To view quotas for a specific user
◆ Use the vxquota command to view quotas for a specific user:
vxquota -v username
Turning off quotas

You can turn off quotas for a mounted file system using the vxquotaoff command.
To turn off quotas for a file system
◆ Turn off quotas for a file system:
vxquotaoff mount_point
228 Quick Reference
Using quotas
Example of turning off quotas

The following example turns off quotas for a VxFS file system mounted at /mnt.
To turn off quotas
◆ Turn off quotas for a VxFS file system mounted at /mnt:
# vxquotaoff /mnt
Appendix B
Diagnostic messages
■ File system response to problems
■ About kernel messages
■ Kernel messages
■ About unique message identifiers
■ Unique message identifiers
File system response to problems

When the file system encounters problems, it responds in one of the following
ways:
Marking an inode bad Inodes can be marked bad if an inode update or a directory-block
update fails. In these types of failures, the file system does not
know what information is on the disk, and considers all the
information that it finds to be invalid. After an inode is marked
bad, the kernel still permits access to the file name, but any
attempt to access the data in the file or change the inode fails.
Disabling transactions If the file system detects an error while writing the intent log, it
disables transactions. After transactions are disabled, the files in
the file system can still be read or written, but no block or inode
frees or allocations, structural changes, directory entry changes,
or other changes to metadata are allowed.
230 Diagnostic messages
About kernel messages
Disabling a file system If an error occurs that compromises the integrity of the file system,
VxFS disables itself. If the intent log fails or an inode-list error
occurs, the super-block is ordinarily updated (setting the
VX_FULLFSCK flag) so that the next fsck does a full structural
check. If this super-block update fails, any further changes to the
file system can cause inconsistencies that are undetectable by the
intent log replay. To avoid this situation, the file system disables
itself.
Recovering a disabled file system

When the file system is disabled, no data can be written to the disk. Although
some minor file system operations still work, most simply return EIO. The only
thing that can be done when the file system is disabled is to do a umount and run
a full fsck.
Warning: Be careful when running this command. By specifying the –y option, all
fsck user prompts are answered with a “yes”, which can make irreversible changes
if it performs a full file system check.
Although a log replay may produce a clean file system, do a full structural check
to be safe.
The file system usually becomes disabled because of disk errors. Disk failures that
disable a file system should be fixed as quickly as possible.
See the fsck_vxfs(1M) manual page.
To execute a full structural check
◆ Use the fsck command to execute a full structural check:
# fsck -F vxfs -o full -y /dev/vx/rdsk/diskgroup/volume
About kernel messages

Kernel messages are diagnostic or error messages generated by the Veritas File
System (VxFS) kernel. Each message has a description and a suggestion on how
to handle or correct the underlying problem.
About global message IDs

When a VxFS kernel message displays on the system console, it is preceded by a
numerical ID shown in the msgcnt field. This ID number increases with each
Diagnostic messages 231
Kernel messages
instance of the message to guarantee that the sequence of events is known when
analyzing file system problems.
Each message is also written to an internal kernel buffer that you can view in the
file /var/adm/messages.
In some cases, additional data is written to the kernel buffer. For example, if an
inode is marked bad, the contents of the bad inode are written. When an error
message is displayed on the console, you can use the unique message ID to find
the message in /var/adm/messages and obtain the additional information.
Kernel messages
Some commonly encountered kernel messages are described on the following
table:
Table B-1 Kernel messages
Message Number Message and Definition
001 NOTICE: msgcnt x: mesg 001: V-2-1: vx_nospace - mount_point file

system full (n block extent)
■ Description
The file system is out of space.
Often, there is plenty of space and one runaway process used up
all the remaining free space. In other cases, the available free space
becomes fragmented and unusable for some files.
■ Action
Monitor the free space in the file system and prevent it from
becoming full. If a runaway process has used up all the space, stop
that process, find the files created by the process, and remove
them. If the file system is out of space, remove files, defragment,
or expand the file system.
To remove files, use the find command to locate the files that are
to be removed. To get the most space with the least amount of
work, remove large files or file trees that are no longer needed. To
defragment or expand the file system, use the fsadm. command.
Kernel messages
Table B-1 Kernel messages (continued)
002 WARNING: msgcnt x: mesg 002: V-2-2: vx_snap_strategy -

mount_point file system write attempt to read-only file system
WARNING: msgcnt x: mesg 002: V-2-2: vx_snap_copyblk - mount_point

file system write attempt to read-only file system
■ Description
The kernel tried to write to a read-only file system. This is an
unlikely problem, but if it occurs, the file system is disabled.
■ Action
The file system was not written, so no action is required. Report
this as a bug to your customer support organization.
003, 004, 005 WARNING: msgcnt x: mesg 003: V-2-3: vx_mapbad - mount_point file
system free extent bitmap in au aun marked bad
WARNING: msgcnt x: mesg 004: V-2-4: vx_mapbad - mount_point file

system free inode bitmap in au aun marked bad
WARNING: msgcnt x: mesg 005: V-2-5: vx_mapbad - mount_point file

system inode extended operation bitmap in au aun marked bad
■ Description
If there is an I/O failure while writing a bitmap, the map is marked
bad. The kernel considers the maps to be invalid, so does not do
any more resource allocation from maps. This situation can cause
the file system to report out of space or out of inode error messages
even though df may report an adequate amount of free space.
This error may also occur due to bitmap inconsistencies. If a bitmap
fails a consistency check, or blocks are freed that are already free
in the bitmap, the file system has been corrupted. This may have
occurred because a user or process wrote directly to the device or
used fsdb to change the file system.
The VX_FULLFSCK flag is set. If the map that failed was a free
extent bitmap, and the VX_FULLFSCK flag can't be set, then the
file system is disabled.
■ Action
Check the console log for I/O errors. If the problem is a disk failure,
replace the disk. If the problem is not related to an I/O failure, find
out how the disk became corrupted. If no user or process was
writing to the device, report the problem to your customer support
organization. Unmount the file system and use fsck to run a full
structural check.
Kernel messages
006, 007 WARNING: msgcnt x: mesg 006: V-2-6: vx_sumupd - mount_point file
system summary update in au aun failed
WARNING: msgcnt x: mesg 007: V-2-7: vx_sumupd - mount_point file

system summary update in inode au iaun failed
■ Description
An I/O error occurred while writing the allocation unit or inode
allocation unit bitmap summary to disk. This sets the
VX_FULLFSCK flag on the file system. If the VX_FULLFSCK flag
can't be set, the file system is disabled.
■ Action
Check the console log for I/O errors. If the problem was caused by
a disk failure, replace the disk before the file system is mounted
for write access, and use fsck to run a full structural check.
008, 009 WARNING: msgcnt x: mesg 008: V-2-8: vx_direrr: function -

mount_point file system dir inode dir_inumber dev/block
device_ID/block dirent inode dirent_inumber error errno
WARNING: msgcnt x: mesg 009: V-2-9: vx_direrr: function -

mount_point file system dir inode dir_inumber dirent inode
dirent_inumber immediate directory error errno
■ Description
A directory operation failed in an unexpected manner. The mount
point, inode, and block number identify the failing directory. If the
inode is an immediate directory, the directory entries are stored
in the inode, so no block number is reported. If the error is ENOENT
or ENOTDIR, an inconsistency was detected in the directory block.
This inconsistency could be a bad free count, a corrupted hash
chain, or any similar directory structure error. If the error is EIO
or ENXIO, an I/O failure occurred while reading or writing the disk
block.
The VX_FULLFSCK flag is set in the super-block so that fsck will
do a full structural check the next time it is run.
■ Action
for write access. Unmount the file system and use fsck to run a full
structural check.
Kernel messages
010 WARNING: msgcnt x: mesg 010: V-2-10: vx_ialloc - mount_point file

system inode inumber not free
■ Description
When the kernel allocates an inode from the free inode bitmap, it
checks the mode and link count of the inode. If either is non-zero,
the free inode bitmap or the inode list is corrupted.
■ Action
Unmount the file system and use fsck to run a full structural check.
011 NOTICE: msgcnt x: mesg 011: V-2-11: vx_noinode - mount_point file

system out of inodes
■ Description
The file system is out of inodes.
■ Action
Monitor the free inodes in the file system. If the file system is
getting full, create more inodes either by removing files or by
expanding the file system.
See the fsadm_vxfs(1M) online manual page.
012 WARNING: msgcnt x: mesg 012: V-2-12: vx_iget - mount_point file

system invalid inode number inumber
■ Description
When the kernel tries to read an inode, it checks the inode number
against the valid range. If the inode number is out of range, the
data structure that referenced the inode number is incorrect and
must be fixed.
■ Action
Kernel messages
013 WARNING: msgcnt x: mesg 013: V-2-13: vx_iposition - mount_point

file system inode inumber invalid inode list extent
■ Description
For a Version 2 and above disk layout, the inode list is dynamically
allocated. When the kernel tries to read an inode, it must look up
the location of the inode in the inode list file. If the kernel finds a
bad extent, the inode can't be accessed. All of the inode list extents
are validated when the file system is mounted, so if the kernel finds
a bad extent, the integrity of the inode list is questionable. This is
a very serious error.
The VX_FULLFSCK flag is set in the super-block and the file system
is disabled.
■ Action
014 WARNING: msgcnt x: mesg 014: V-2-14: vx_iget - inode table overflow
■ Description
All the system in-memory inodes are busy and an attempt was
made to use a new inode.
■ Action
Note: Note: The tunable parameter vx_ninode is used to set the

value of vxfs_ninode.
Look at the processes that are running and determine which
processes are using inodes. If it appears there are runaway
processes, they might be tying up the inodes. If the system load
appears normal, increase the vxfs_ninode parameter in the kernel.
Look at the processes that are running and determine which
processes are using inodes. If it appears there are runaway
processes, they might be tying up the inodes. If the system load
appears normal, increase the vx_ninode parameter in the kernel.
See “Using kernel tunables” on page 40.
Kernel messages
015 WARNING: msgcnt x: mesg 015: V-2-15: vx_ibadinactive - mount_point

file system can't mark inode inumber bad
WARNING: msgcnt x: mesg 015: V-2-15: vx_ilisterr - mount_point file

system can't mark inode inumber bad
■ Description
An attempt to mark an inode bad on disk, and the super-block
update to set the VX_FULLFSCK flag, failed. This indicates that a
catastrophic disk error may have occurred since both an inode list
block and the super-block had I/O failures. The file system is
disabled to preserve file system integrity.
■ Action
Check the console log for I/O errors. If the disk failed, replace it
before remounting the file system.
016 WARNING: msgcnt x: mesg 016: V-2-16: vx_ilisterr - mount_point file

system error reading inode inumber
■ Description
An I/O error occurred while reading the inode list. The
VX_FULLFSCK flag is set.
■ Action
for write access. Unmount the file system and use fsck to run a full
structural check.
Kernel messages
017
Kernel messages
WARNING: msgcnt x: mesg 017: V-2-17: vx_attr_getblk - mount_point

file system inode inumber marked bad in core
WARNING: msgcnt x: mesg 017: V-2-17: vx_attr_iget - mount_point

WARNING: msgcnt x: mesg 017: V-2-17: vx_attr_indadd - mount_point

WARNING: msgcnt x: mesg 017: V-2-17: vx_attr_indtrunc -

mount_point file system inode inumber marked bad in core
WARNING: msgcnt x: mesg 017: V-2-17: vx_attr_iremove -

WARNING: msgcnt x: mesg 017: V-2-17: vx_bmap - mount_point file

system inode inumber marked bad in core
WARNING: msgcnt x: mesg 017: V-2-17: vx_bmap_indirect_ext4 -

WARNING: msgcnt x: mesg 017: V-2-17: vx_delbuf_flush - mount_point

WARNING: msgcnt x: mesg 017: V-2-17: vx_dio_iovec - mount_point

WARNING: msgcnt x: mesg 017: V-2-17: vx_dirbread - mount_point

WARNING: msgcnt x: mesg 017: V-2-17: vx_dircreate - mount_point

WARNING: msgcnt x: mesg 017: V-2-17: vx_dirlook - mount_point file

WARNING: msgcnt x: mesg 017: V-2-17: vx_doextop_iau - mount_point

WARNING: msgcnt x: mesg 017: V-2-17: vx_doextop_now -

WARNING: msgcnt x: mesg 017: V-2-17: vx_do_getpage - mount_point

WARNING: msgcnt x: mesg 017: V-2-17: vx_enter_ext4 - mount_point

WARNING: msgcnt x: mesg 017: V-2-17: vx_exttrunc - mount_point

WARNING: msgcnt x: mesg 017: V-2-17: vx_get_alloc - mount_point

Kernel messages
017 (continued) WARNING: msgcnt x: mesg 017: V-2-17: vx_ilisterr - mount_point file
WARNING: msgcnt x: mesg 017: V-2-17: vx_indtrunc - mount_point

WARNING: msgcnt x: mesg 017: V-2-17: vx_iread - mount_point file

WARNING: msgcnt x: mesg 017: V-2-17: vx_iremove - mount_point

WARNING: msgcnt x: mesg 017: V-2-17: vx_iremove_attr -

WARNING: msgcnt x: mesg 017: V-2-17: vx_logwrite_flush -

WARNING: msgcnt x: mesg 017: V-2-17: vx_oltmount_iget -

WARNING: msgcnt x: mesg 017: V-2-17: vx_overlay_bmap -

WARNING: msgcnt x: mesg 017: V-2-17: vx_readnomap - mount_point

WARNING: msgcnt x: mesg 017: V-2-17: vx_reorg_trunc - mount_point

WARNING: msgcnt x: mesg 017: V-2-17: vx_stablestore - mount_point

WARNING: msgcnt x: mesg 017: V-2-17: vx_tranitimes - mount_point

WARNING: msgcnt x: mesg 017: V-2-17: vx_trunc - mount_point file

WARNING: msgcnt x: mesg 017: V-2-17: vx_write_alloc2 - mount_point

WARNING: msgcnt x: mesg 017: V-2-17: vx_write_default -

WARNING: msgcnt x: mesg 017: V-2-17: vx_zero_alloc - mount_point

Kernel messages
017 (continued) ■ Description

When inode information is no longer dependable, the kernel marks
it bad in memory. This is followed by a message to mark it bad on
disk as well unless the mount command ioerror option is set to
disable, or there is subsequent I/O failure when updating the inode
on disk. No further operations can be performed on the inode.
The most common reason for marking an inode bad is a disk I/O
failure. If there is an I/O failure in the inode list, on a directory
block, or an indirect address extent, the integrity of the data in the
inode, or the data the kernel tried to write to the inode list, is
questionable. In these cases, the disk driver prints an error message
and one or more inodes are marked bad.
The kernel also marks an inode bad if it finds a bad extent address,
invalid inode fields, or corruption in directory data blocks during
a validation check. A validation check failure indicates the file
system has been corrupted. This usually occurs because a user or
process has written directly to the device or used fsdb to change
the file system.
The VX_FULLFSCK flag is set in the super-block so fsck will do a
full structural check the next time it is run.
■ Action
out how the disk became corrupted. If no user or process is writing
to the device, report the problem to your customer support
organization. In either case, unmount the file system. The file
system can be remounted without a full fsck unless the
VX_FULLFSCK flag is set for the file system.
019 WARNING: msgcnt x: mesg 019: V-2-19: vx_log_add - mount_point

file system log overflow
■ Description
Log ID overflow. When the log ID reaches VX_MAXLOGID
(approximately one billion by default), a flag is set so the file system
resets the log ID at the next opportunity. If the log ID has not been
reset, when the log ID reaches VX_DISLOGID (approximately
VX_MAXLOGID plus 500 million by default), the file system is
disabled. Since a log reset will occur at the next 60 second sync
interval, this should never happen.
■ Action
Kernel messages
020 WARNING: msgcnt x: mesg 020: V-2-20: vx_logerr - mount_point file

system log error errno
■ Description
Intent log failed. The kernel will try to set the VX_FULLFSCK and
VX_LOGBAD flags in the super-block to prevent running a log
replay. If the super-block can't be updated, the file system is
disabled.
■ Action
Check the console log for I/O errors. If the disk failed, replace it
before remounting the file system.
021 WARNING: msgcnt x: mesg 021: V-2-21: vx_fs_init - mount_point file

system validation failure
■ Description
When a VxFS file system is mounted, the structure is read from
disk. If the file system is marked clean, the structure is correct and
the first block of the intent log is cleared.
If there is any I/O problem or the structure is inconsistent, the
kernel sets the VX_FULLFSCK flag and the mount fails.
If the error isn't related to an I/O failure, this may have occurred
because a user or process has written directly to the device or used
fsdb to change the file system.
■ Action
organization. In either case, unmount the file system and use fsck
to run a full structural check.
Kernel messages
022 WARNING: msgcnt x: mesg 022: V-2-22: vx_mountroot - root file

system remount failed
■ Description
The remount of the root file system failed. The system will not be
usable if the root file system can't be remounted for read/write
access.
When a root Veritas File System is first mounted, it is mounted
for read-only access. After fsck is run, the file system is remounted
for read/write access. The remount fails if fsck completed a resize
operation or modified a file that was opened before the fsck was
run. It also fails if an I/O error occurred during the remount.
Usually, the system halts or reboots automatically.
■ Action
Reboot the system. The system either remounts the root cleanly
or runs a full structural fsck and remounts cleanly. If the remount
succeeds, no further action is necessary.
Check the console log for I/O errors. If the disk has failed, replace
it before the file system is mounted for write access.
If the system won't come up and a full structural fsck hasn't been
run, reboot the system on a backup root and manually run a full
structural fsck. If the problem persists after the full structural fsck
and there are no I/O errors, contact your customer support
organization.
023 WARNING: msgcnt x: mesg 023: V-2-23: vx_unmountroot - root file

system is busy and can't be unmounted cleanly
■ Description
There were active files in the file system and they caused the
unmount to fail.
When the system is halted, the root file system is unmounted. This
happens occasionally when a process is hung and it can't be killed
before unmounting the root.
■ Action
fsck will run when the system is rebooted. It should clean up the
file system. No other action is necessary.
If the problem occurs every time the system is halted, determine
the cause and contact your customer support organization.
Kernel messages
024 WARNING: msgcnt x: mesg 024: V-2-24: vx_cutwait - mount_point

file system current usage table update error
■ Description
Update to the current usage table (CUT) failed.
For a Version 2 disk layout, the CUT contains a fileset version
number and total number of blocks used by each fileset.
The VX_FULLFSCK flag is set in the super-block. If the super-block
can't be written, the file system is disabled.
■ Action
025 WARNING: msgcnt x: mesg 025: V-2-25: vx_wsuper - mount_point

file system super-block update failed
■ Description
An I/O error occurred while writing the super-block during a resize
operation. The file system is disabled.
■ Action
replace the disk before the file system is mounted for write access.
026 WARNING: msgcnt x: mesg 026: V-2-26: vx_snap_copyblk -

mount_point primary file system read error
■ Description
Snapshot file system error.
When the primary file system is written, copies of the original data
must be written to the snapshot file system. If a read error occurs
on a primary file system during the copy, any snapshot file system
that doesn't already have a copy of the data is out of date and must
be disabled.
■ Action
An error message for the primary file system prints. Resolve the
error on the primary file system and rerun any backups or other
applications that were using the snapshot that failed when the
error occurred.
Kernel messages
027 WARNING: msgcnt x: mesg 027: V-2-27: vx_snap_bpcopy -

mount_point snapshot file system write error
■ Description
A write to the snapshot file system failed.
As the primary file system is updated, copies of the original data
are read from the primary file system and written to the snapshot
file system. If one of these writes fails, the snapshot file system is
disabled.
■ Action
it. Resolve the error on the disk and rerun any backups or other
applications that were using the snapshot that failed when the
error occurred.
028 WARNING: msgcnt x: mesg 028: V-2-28: vx_snap_alloc - mount_point

snapshot file system out of space
■ Description
The snapshot file system ran out of space to store changes.
During a snapshot backup, as the primary file system is modified,
the original data is copied to the snapshot file system. This error
can occur if the snapshot file system is left mounted by mistake,
if the snapshot file system was given too little disk space, or the
primary file system had an unexpected burst of activity. The
snapshot file system is disabled.
■ Action
Make sure the snapshot file system was given the correct amount
of space. If it was, determine the activity level on the primary file
system. If the primary file system was unusually busy, rerun the
backup. If the primary file system is no busier than normal, move
the backup to a time when the primary file system is relatively idle
or increase the amount of disk space allocated to the snapshot file
system.
Rerun any backups that failed when the error occurred.
Kernel messages
029, 030 WARNING: msgcnt x: mesg 029: V-2-29: vx_snap_getbp - mount_point

snapshot file system block map write error
WARNING: msgcnt x: mesg 030: V-2-30: vx_snap_getbp - mount_point

snapshot file system block map read error
■ Description
During a snapshot backup, each snapshot file system maintains a
block map on disk. The block map tells the snapshot file system
where data from the primary file system is stored in the snapshot
file system. If an I/O operation to the block map fails, the snapshot
file system is disabled.
■ Action
it. Resolve the error on the disk and rerun any backups that failed
when the error occurred.
031 WARNING: msgcnt x: mesg 031: V-2-31: vx_disable - mount_point file

system disabled
■ Description
File system disabled, preceded by a message that specifies the
reason. This usually indicates a serious disk problem.
■ Action
If the problem is a disk failure, replace the disk before the file
system is mounted for write access.
032 WARNING: msgcnt x: mesg 032: V-2-32: vx_disable - mount_point

snapshot file system disabled
■ Description
Snapshot file system disabled, preceded by a message that specifies
the reason.
■ Action
Unmount the snapshot file system, correct the problem specified
by the message, and rerun any backups that failed due to the error.
Kernel messages
033 WARNING: msgcnt x: mesg 033: V-2-33: vx_check_badblock -

mount_point file system had an I/O error, setting VX_FULLFSCK
■ Description
When the disk driver encounters an I/O error, it sets a flag in the
super-block structure. If the flag is set, the kernel will set the
VX_FULLFSCK flag as a precautionary measure. Since no other
error has set the VX_FULLFSCK flag, the failure probably occurred
on a data block.
■ Action
034 WARNING: msgcnt x: mesg 034: V-2-34: vx_resetlog - mount_point

file system can't reset log
■ Description
The kernel encountered an error while resetting the log ID on the
file system. This happens only if the super-block update or log
write encountered a device failure. The file system is disabled to
preserve its integrity.
■ Action
035 WARNING: msgcnt x: mesg 035: V-2-35: vx_inactive - mount_point

file system inactive of locked inode inumber
■ Description
VOP_INACTIVE was called for an inode while the inode was being
used. This should never happen, but if it does, the file system is
disabled.
■ Action
Report as a bug to your customer support organization.
Kernel messages
036 WARNING: msgcnt x: mesg 036: V-2-36: vx_lctbad - mount_point file

system link count table lctnumber bad
■ Description
Update to the link count table (LCT) failed.
For a Version 2 and above disk layout, the LCT contains the link
count for all the structural inodes. The VX_FULLFSCK flag is set
in the super-block. If the super-block can't be written, the file
system is disabled.
■ Action
037 WARNING: msgcnt x: mesg 037: V-2-37: vx_metaioerr - function -

volume_name file system meta data [read|write] error in dev/block
device_ID/block
■ Description
A read or a write error occurred while accessing file system
metadata. The full fsck flag on the file system was set. The message
specifies whether the disk I/O that failed was a read or a write.
File system metadata includes inodes, directory blocks, and the
file system log. If the error was a write error, it is likely that some
data was lost. This message should be accompanied by another file
system message describing the particular file system metadata
affected, as well as a message from the disk driver containing
information about the disk I/O error.
■ Action
Resolve the condition causing the disk error. If the error was the
result of a temporary condition (such as accidentally turning off
a disk or a loose cable), correct the condition. Check for loose cables,
etc. Unmount the file system and use fsck to run a full structural
check (possibly with loss of data).
In case of an actual disk error, if it was a read error and the disk
driver remaps bad sectors on write, it may be fixed when fsck is
run since fsck is likely to rewrite the sector with the read error. In
other cases, you replace or reformat the disk drive and restore the
file system from backups. Consult the documentation specific to
your system for information on how to recover from disk errors.
The disk driver should have printed a message that may provide
more information.
Kernel messages
038 WARNING: msgcnt x: mesg 038: V-2-38: vx_dataioerr - volume_name

file system file data [read|write] error in dev/block device_ID/block
■ Description
A read or a write error occurred while accessing file data. The
message specifies whether the disk I/O that failed was a read or a
write. File data includes data currently in files and free blocks. If
the message is printed because of a read or write error to a file,
another message that includes the inode number of the file will
print. The message may be printed as the result of a read or write
error to a free block, since some operations allocate an extent and
immediately perform I/O to it. If the I/O fails, the extent is freed
and the operation fails. The message is accompanied by a message
from the disk driver regarding the disk I/O error.
■ Action
etc. If any file data was lost, restore the files from backups.
Determine the file names from the inode number.
See the ncheck(1M) manual page.
If an actual disk error occurred, make a backup of the file system,
replace or reformat the disk drive, and restore the file system from
the backup. Consult the documentation specific to your system for
information on how to recover from disk errors. The disk driver
should have printed a message that may provide more information.
Kernel messages
039 WARNING: msgcnt x: mesg 039: V-2-39: vx_writesuper - file system

super-block write error
■ Description
An attempt to write the file system super block failed due to a disk
I/O error. If the file system was being mounted at the time, the
mount will fail. If the file system was mounted at the time and the
full fsck flag was being set, the file system will probably be disabled
and Message 031 will also be printed. If the super-block was being
written as a result of a sync operation, no other action is taken.
■ Action
etc. Unmount the file system and use fsck to run a full structural
check.
If an actual disk error occurred, make a backup of the file system,
replace or reformat the disk drive, and restore the file system from
backups. Consult the documentation specific to your system for
information on how to recover from disk errors. The disk driver
should have printed a message that may provide more information.
040 WARNING: msgcnt x: mesg 040: V-2-40: vx_dqbad - mount_point file

system user|group quota file update error for id id
■ Description
An update to the user quotas file failed for the user ID.
The quotas file keeps track of the total number of blocks and inodes
used by each user, and also contains soft and hard limits for each
user ID. The VX_FULLFSCK flag is set in the super-block. If the
super-block cannot be written, the file system is disabled.
■ Action
Check the console log for I/O errors. If the disk has a hardware
failure, it should be repaired before the file system is mounted for
write access.
Kernel messages
041 WARNING: msgcnt x: mesg 041: V-2-41: vx_dqget - mount_point file

system user|group quota file can't read quota for id id
■ Description
A read of the user quotas file failed for the uid.
The quotas file keeps track of the total number of blocks and inodes
used by each user, and contains soft and hard limits for each user
ID. The VX_FULLFSCK flag is set in the super-block. If the
super-block cannot be written, the file system is disabled.
■ Action
Check the console log for I/O errors. If the disk has a hardware
failure, it should be repaired before the file system is mounted for
write access.
042 WARNING: msgcnt x: mesg 042: V-2-42: vx_bsdquotaupdate -

mount_point file system user|group_id disk limit reached
■ Description
The hard limit on blocks was reached. Further attempts to allocate
blocks for files owned by the user will fail.
■ Action
Remove some files to free up space.

mount_point file system user|group_id disk quota exceeded too long
■ Description
The soft limit on blocks was exceeded continuously for longer than
the soft quota time limit. Further attempts to allocate blocks for
files will fail.
■ Action

mount_point file system user|group_id disk quota exceeded
■ Description
The soft limit on blocks is exceeded. Users can exceed the soft limit
for a limited amount of time before allocations begin to fail. After
the soft quota time limit has expired, subsequent attempts to
allocate blocks for files fail.
■ Action
Kernel messages
045 WARNING: msgcnt x: mesg 045: V-2-45: vx_bsdiquotaupdate -

mount_point file system user|group_id inode limit reached
■ Description
The hard limit on inodes was exceeded. Further attempts to create
files owned by the user will fail.
■ Action
Remove some files to free inodes.
046 WARNING: msgcnt x: mesg 046: V-2-46: vx_bsdiquotaupdate -

mount_point file system user|group_id inode quota exceeded too long
■ Description
The soft limit on inodes has been exceeded continuously for longer
than the soft quota time limit. Further attempts to create files
owned by the user will fail.
■ Action
047 WARNING: msgcnt x: mesg 047: V-2-47: vx_bsdiquotaupdate - warning:

mount_point file system user|group_id inode quota exceeded
■ Description
The soft limit on inodes was exceeded. The soft limit can be
exceeded for a certain amount of time before attempts to create
new files begin to fail. Once the time limit has expired, further
attempts to create files owned by the user will fail.
■ Action
Kernel messages
048, 049 WARNING: msgcnt x: mesg 048: V-2-48: vx_dqread - warning:

mount_point file system external user|group quota file read failed
WARNING: msgcnt x: mesg 049: V-2-49: vx_dqwrite - warning:

mount_point file system external user|group quota file write failed
■ Description
To maintain reliable usage counts, VxFS maintains the user quotas
file as a structural file in the structural fileset.
These files are updated as part of the transactions that allocate
and free blocks and inodes. For compatibility with the quota
administration utilities, VxFS also supports the standard user
visible quota files.
When quotas are turned off, synced, or new limits are added, VxFS
tries to update the external quota files. When quotas are enabled,
VxFS tries to read the quota limits from the external quotas file.
If these reads or writes fail, the external quotas file is out of date.
■ Action
Determine the reason for the failure on the external quotas file
and correct it. Recreate the quotas file.
Kernel messages
056 WARNING: msgcnt x: mesg 056: V-2-56: vx_mapbad - mount_point

file system extent allocation unit state bitmap number number marked
bad
■ Description
If there is an I/O failure while writing a bitmap, the map is marked
bad. The kernel considers the maps to be invalid, so does not do
any more resource allocation from maps. This situation can cause
the file system to report “out of space” or “out of inode” error
messages even though df may report an adequate amount of free
space.
This error may also occur due to bitmap inconsistencies. If a bitmap
fails a consistency check, or blocks are freed that are already free
in the bitmap, the file system has been corrupted. This may have
occurred because a user or process wrote directly to the device or
used fsdb to change the file system.
The VX_FULLFSCK flag is set. If the VX_FULLFSCK flag can't be
set, the file system is disabled.
■ Action
structural check.
057 WARNING: msgcnt x: mesg 057: V-2-57: vx_esum_bad - mount_point

file system extent allocation unit summary number number marked
bad
■ Description
An I/O error occurred reading or writing an extent allocation unit
summary.
The VX_FULLFSCK flag is set. If the VX_FULLFSCK flag can't be
■ Action
structural check.
Kernel messages
058 WARNING: msgcnt x: mesg 058: V-2-58: vx_isum_bad - mount_point

file system inode allocation unit summary number number marked
bad
■ Description
An I/O error occurred reading or writing an inode allocation unit
summary.
The VX_FULLFSCK flag is set. If the VX_FULLFSCK flag cannot be
■ Action
structural check.
059 WARNING: msgcnt x: mesg 059: V-2-59: vx_snap_getbitbp -

mount_point snapshot file system bitmap write error
■ Description
An I/O error occurred while writing to the snapshot file system
bitmap. There is no problem with the snapped file system, but the
■ Action
organization. Restart the snapshot on an error free disk partition.
Kernel messages
060 WARNING: msgcnt x: mesg 060: V-2-60: vx_snap_getbitbp -

mount_point snapshot file system bitmap read error
■ Description
An I/O error occurred while reading the snapshot file system
bitmap. There is no problem with snapped file system, but the
■ Action
organization. Restart the snapshot on an error free disk partition.
061 WARNING: msgcnt x: mesg 061: V-2-61: vx_resize - mount_point file

system remount failed
■ Description
During a file system resize, the remount to the new size failed. The
VX_FULLFSCK flag is set and the file system is disabled.
■ Action
After the check, the file system shows the new size.
062 NOTICE: msgcnt x: mesg 062: V-2-62: vx_attr_creatop - invalid

disposition returned by attribute driver
■ Description
A registered extended attribute intervention routine returned an
invalid return code to the VxFS driver during extended attribute
inheritance.
■ Action
Determine which vendor supplied the registered extended attribute
intervention routine and contact their customer support
organization.
Kernel messages
063 WARNING: msgcnt x: mesg 063: V-2-63: vx_fset_markbad -

mount_point file system mount_point fileset (index number) marked
bad
■ Description
An error occurred while reading or writing a fileset structure.
VX_FULLFSCK flag is set. If the VX_FULLFSCK flag can't be set,
the file system is disabled.
■ Action
064 WARNING: msgcnt x: mesg 064: V-2-64: vx_ivalidate - mount_point

file system inode number version number exceeds fileset's
■ Description
During inode validation, a discrepancy was found between the
inode version number and the fileset version number. The inode
may be marked bad, or the fileset version number may be changed,
depending on the ratio of the mismatched version numbers.
VX_FULLFSCK flag is set. If the VX_FULLFSCK flag can't be set,
the file system is disabled.
■ Action
066 NOTICE: msgcnt x: mesg 066: V-2-66: DMAPI mount event - buffer
■ Description
An HSM (Hierarchical Storage Management) agent responded to
a DMAPI mount event and returned a message in buffer.
■ Action
Consult the HSM product documentation for the appropriate
response to the message.
Kernel messages
067 WARNING: msgcnt x: mesg 067: V-2-67: mount of device_path requires

HSM agent
■ Description
The file system mount failed because the file system was marked
as being under the management of an HSM agent, and no HSM
agent was found during the mount.
■ Action
Restart the HSM agent and try to mount the file system again.
068 WARNING: msgcnt x: mesg 068: V-2-68: ncsize parameter is greater

than 80% of the vxfs_ninode parameter; increasing the value of
vxfs:vxfs_ninode
■ Description
The value auto-tuned for the vxfs_ninode parameter is less than
125% of the ncsize parameter.
■ Action
To prevent this message from occurring, set vxfs_ninode to at least
125% of the value of ncsize. The best way to do this is to adjust
ncsize down, rather than adjusting vxfs_ninode up.
Kernel messages
069 WARNING: msgcnt x: mesg 069: V-2-69: memory usage specified by

the vxfs:vxfs_ninode and vxfs:vx_bc_bufhwm parameters exceeds
available memory; the system may hang under heavy load
■ Description
The value of the system tunable parameters—vxfs_ninode and
vx_bc_bufhwm—add up to a value that is more than 66% of the
kernel virtual address space or more than 50% of the physical
system memory. VxFS inodes require approximately one kilobyte
each, so both values can be treated as if they are in units of one
kilobyte.
The value of the system tunable parameters—vx_ninode and
vx_bc_bufhwm—add up to a value that is more than 66% of the
kernel virtual address space or more than 50% of the physical
system memory. VxFS inodes require approximately one kilobyte
each, so both values can be treated as if they are in units of one
kilobyte.
■ Action
To avoid a system hang, reduce the value of one or both parameters
to less than 50% of physical memory or to 66% of kernel virtual
memory.
Note: The tunable parameter vx_ninode is used to set the value

of vxfs_ninode.
070 WARNING: msgcnt x: mesg 070: V-2-70: checkpoint checkpoint_name

removed from file system mount_point
■ Description
The file system ran out of space while updating a Storage
Checkpoint. The Storage Checkpoint was removed to allow the
operation to complete.
■ Action
Increase the size of the file system. If the file system size cannot
be increased, remove files to create sufficient space for new Storage
Checkpoints. Monitor capacity of the file system closely to ensure
it does not run out of space.
Kernel messages
071 NOTICE: msgcnt x: mesg 071: V-2-71: cleared data I/O error flag in
mount_point file system
■ Description
The user data I/O error flag was reset when the file system was
mounted. This message indicates that a read or write error occurred
while the file system was previously mounted.
See Message Number 038.
■ Action
Informational only, no action required.
072 WARNING: msgcnt x: vxfs: mesg 072: could not failover for
volume_name file system
■ Description
This message is specific to the cluster file system. The message
indicates a problem in a scenario where a node failure has occurred
in the cluster and the newly selected primary node encounters a
failure.
■ Action
Save the system logs and core dump of the node along with the
disk image (metasave) and contact your customer support
organization. The node can be rebooted to join the cluster.
074 WARNING: msgcnt x: mesg 074: V-2-74: vx_rcfg - fileset_number

fileset not found while remounting mount_point
■ Description
This message is specific to the cluster file system. The message
indicates a problem in a scenario where a node failure has occurred
in the cluster and the newly selected primary node encounters a
failure.
■ Action
Save the core dump of the node and contact your customer support
organization. The node can be rebooted to join the cluster.
Kernel messages
075 WARNING: msgcnt x: mesg 075: V-2-75: replay fsck failed for
mount_point file system
■ Description
The log replay failed during a failover or while migrating the CFS
primary-ship to one of the secondary cluster nodes. The file system
was disabled.
■ Action
Unmount the file system from the cluster. Use fsck to run a full
structural check and mount the file system again.
076 NOTICE: msgcnt x: mesg 076: V-2-76: checkpoint asynchronous

operation on mount_point file system still in progress
■ Description
An EBUSY message was received while trying to unmount a file system.

The unmount failure was caused by a pending asynchronous fileset
operation, such as a fileset removal or fileset conversion to a nodata
Storage Checkpoint.
■ Action
The operation may take a considerable length of time. You can do
a forced unmount, or simply wait for the operation to complete so
file system can be unmounted cleanly.
The operation may take a considerable length of time. Wait for the
operation to complete so file system can be unmounted cleanly.
Kernel messages
077 WARNING: msgcnt x: mesg 077: V-2-77: vx_fshdchange - mount_point

file system number fileset, fileset header: checksum failed
■ Description
Disk corruption was detected while changing fileset headers. This
can occur when writing a new inode allocation unit, preventing
the allocation of new inodes in the fileset.
■ Action
078 WARNING: msgcnt x: mesg 078: V-2-78: vx_ilealloc - mount_point file

system mount_point fileset (index number) ilist corrupt
■ Description
The inode list for the fileset was corrupted and the corruption was
detected while allocating new inodes. The failed system call returns
an ENOSPC error. Any subsequent inode allocations will fail unless
a sufficient number of files are removed.
■ Action
Kernel messages
079
Kernel messages
WARNING: msgcnt x: mesg 017: V-2-79: vx_attr_getblk - mount_point

file system inode inumber marked bad on disk
WARNING: msgcnt x: mesg 017: V-2-79: vx_attr_iget - mount_point

WARNING: msgcnt x: mesg 017: V-2-79: vx_attr_indadd - mount_point

WARNING: msgcnt x: mesg 017: V-2-79: vx_attr_indtrunc -

mount_point file system inode inumber marked bad on disk
WARNING: msgcnt x: mesg 017: V-2-79: vx_attr_iremove -

WARNING: msgcnt x: mesg 017: V-2-79: vx_bmap - mount_point file

system inode inumber marked bad on disk
WARNING: msgcnt x: mesg 017: V-2-79: vx_bmap_indirect_ext4 -

WARNING: msgcnt x: mesg 017: V-2-79: vx_delbuf_flush - mount_point

WARNING: msgcnt x: mesg 017: V-2-79: vx_dio_iovec - mount_point

WARNING: msgcnt x: mesg 017: V-2-79: vx_dirbread - mount_point

WARNING: msgcnt x: mesg 017: V-2-79: vx_dircreate - mount_point

WARNING: msgcnt x: mesg 017: V-2-79: vx_dirlook - mount_point file

WARNING: msgcnt x: mesg 017: V-2-79: vx_doextop_iau - mount_point

WARNING: msgcnt x: mesg 017: V-2-79: vx_doextop_now -

WARNING: msgcnt x: mesg 017: V-2-79: vx_do_getpage - mount_point

WARNING: msgcnt x: mesg 017: V-2-79: vx_enter_ext4 - mount_point

WARNING: msgcnt x: mesg 017: V-2-79: vx_exttrunc - mount_point

WARNING: msgcnt x: mesg 017: V-2-79: vx_get_alloc - mount_point

Kernel messages
079 (continued) WARNING: msgcnt x: mesg 017: V-2-79: vx_ilisterr - mount_point file
WARNING: msgcnt x: mesg 017: V-2-79: vx_indtrunc - mount_point

WARNING: msgcnt x: mesg 017: V-2-79: vx_iread - mount_point file

WARNING: msgcnt x: mesg 017: V-2-79: vx_iremove - mount_point

WARNING: msgcnt x: mesg 017: V-2-79: vx_iremove_attr -

WARNING: msgcnt x: mesg 017: V-2-79: vx_logwrite_flush -

WARNING: msgcnt x: mesg 017: V-2-79: vx_oltmount_iget -

WARNING: msgcnt x: mesg 017: V-2-79: vx_overlay_bmap -

WARNING: msgcnt x: mesg 017: V-2-79: vx_readnomap - mount_point

WARNING: msgcnt x: mesg 017: V-2-79: vx_reorg_trunc - mount_point

WARNING: msgcnt x: mesg 017: V-2-79: vx_stablestore - mount_point

WARNING: msgcnt x: mesg 017: V-2-79: vx_tranitimes - mount_point

WARNING: msgcnt x: mesg 017: V-2-79: vx_trunc - mount_point file

WARNING: msgcnt x: mesg 017: V-2-79: vx_write_alloc2 - mount_point

WARNING: msgcnt x: mesg 017: V-2-79: vx_write_default -

WARNING: msgcnt x: mesg 017: V-2-79: vx_zero_alloc - mount_point

Kernel messages
079 (continued) ■ Description

When inode information is no longer dependable, the kernel marks
it bad on disk. The most common reason for marking an inode bad
is a disk I/O failure. If there is an I/O failure in the inode list, on a
directory block, or an indirect address extent, the integrity of the
data in the inode, or the data the kernel tried to write to the inode
list, is questionable. In these cases, the disk driver prints an error
message and one or more inodes are marked bad.
The kernel also marks an inode bad if it finds a bad extent address,
invalid inode fields, or corruption in directory data blocks during
a validation check. A validation check failure indicates the file
system has been corrupted. This usually occurs because a user or
process has written directly to the device or used fsdb to change
the file system.
The VX_FULLFSCK flag is set in the super-block so fsck will do a
full structural check the next time it is run.
■ Action
Kernel messages
080 WARNING: msgcnt x: mesg 080: V-2-80: Disk layout versions older
than Version 4 will not be supported in the next release. It is advisable
to upgrade to the latest disk layout version now.
See the vxupgrade(1M) manual page.
See the Storage Foundation Release Notes.
■ Action
Use the vxupgrade command to begin upgrading file systems
using older disk layouts to Version 5. Consider the following when
planning disk layout upgrades:
■ Version 1 disk layout file systems can support more than 8 million
inodes, while Version 2 disk layout file systems have an 8 million
inode limit.
The Version 1 disk layout provides finer control of disk geometry
than subsequent disk layouts. This finer control is not relevant on
disks employing newer technologies, but can still be applicable on
older hardware. If you are using Version 1 disk layout file systems
on older hardware that needs fine control of disk geometry, a disk
layout upgrade may be problematic.
Images of Version 1 or Version 2 disk layout file systems created
by copy utilities, such as dd or volcopy, will become unusable
after a disk layout upgrade. Offline conversions tools will be
provided in the next VxFS feature release to aid in migrating
volume-image backup copies of Version 1 and Version 2 disk layout
file systems to a Version 4 disk layout.
081 WARNING: msgcnt x: mesg 081: V-2-81: possible network partition

detected
■ Description
This message displays when CFS detects a possible network
partition and disables the file system locally, that is, on the node
where the message appears.
■ Action
There are one or more private network links for communication
between the nodes in a cluster. At least one link must be active to
maintain the integrity of the cluster. If all the links go down, after
the last network link is broken, the node can no longer
communicate with other nodes in the cluster.
Check the network connections. After verifying that the network
connections is operating correctly, unmount the disabled file
system and mount it again.
Kernel messages
082 WARNING: msgcnt x: mesg 082: V-2-82: volume_name file system is

on shared volume. It may get damaged if cluster is in partitioned state.
■ Description
If a cluster node is in a partitioned state, and if the file system is
on a shared VxVM volume, this volume may become corrupted by
accidental access from another node in the cluster.
■ Action
These shared disks can also be seen by nodes in a different
partition, so they can inadvertently be corrupted. So the second
message 082 tells that the device mentioned is on shared volume
and damage can happen only if it is a real partition problem. Do
not use it on any other node until the file system is unmounted
from the mounted nodes.
083 WARNING: msgcnt x: mesg 083: V-2-83: mount_point file system log
is not compatible with the specified intent log I/O size
■ Description
Either the specified mount logiosize size is not compatible with
the file system layout, or the file system is corrupted.
■ Action
Mount the file system again without specifying the logiosize option,
or use a logiosize value compatible with the intent log specified
when the file system was created. If the error persists, unmount
the file system and use fsck to run a full structural check.
084 WARNING: msgcnt x: mesg 084: V-2-84: in volume_name quota on

failed during assumption. (stage stage_number)
■ Description
In a cluster file system, when the primary of the file system fails,
a secondary file system is chosen to assume the role of the primary.
The assuming node will be able to enforce quotas after becoming
the primary.
If the new primary is unable to enforce quotas this message will
be displayed.
■ Action
Issue the quotaon command from any of the nodes that have the
file system mounted.
Kernel messages
085 WARNING: msgcnt x: mesg 085: V-2-85: Checkpoint quota - warning:

file_system file system fileset quota hard limit exceeded
■ Description
The system administrator sets the quotas for Storage Checkpoints
in the form of a soft limit and hard limit. This message displays
when the hard limit is exceeded.
■ Action
Delete Storage Checkpoints or increase the hard limit.
086 WARNING: msgcnt x: mesg 086: V-2-86: Checkpoint quota - warning:

file_system file system fileset quota soft limit exceeded
■ Description
The system administrator sets the quotas for Storage Checkpoints
in the form of a soft limit and hard limit. This message displays
when the soft limit is exceeded.
■ Action
Delete Storage Checkpoints or increase the soft limit. This is not
a mandatory action, but is recommended.
087 WARNING: msgcnt x: mesg 087: V-2-87: vx_dotdot_manipulate:

file_system file system inumber inode ddnumber dotdot inode error
■ Description
When performing an operation that changes an inode entry, if the
inode is incorrect, this message will display.
■ Action
Run a full file system check using fsck to correct the errors.
088 WARNING: msgcnt x: mesg 088: V-2-88: quotaon on file_system failed;

limits exceed limit
■ Description
The external quota file, quotas, contains the quota values, which
range from 0 up to 2147483647. When quotas are turned on by the
quotaon command, this message displays when a user exceeds
the quota limit.
■ Action
Correct the quota values in the quotas file.
Kernel messages
089 WARNING: msgcnt x: mesg 089: V-2-89: quotaon on file_system

invalid; disk usage for group/user id uid exceeds sectors sectors
■ Description
The supported quota limit is up to 2147483647 sectors. When
quotas are turned on by the quotaon command, this message
displays when a user exceeds the supported quota limit.
■ Action
Ask the user to delete files to lower the quota below the limit.
090 WARNING: msgcnt x: mesg 090: V-2-90: quota on file_system failed;

soft limits greater than hard limits
■ Description
One or more users or groups has a soft limit set greater than the
hard limit, preventing the BSD quota from being turned on.
■ Action
Check the soft limit and hard limit for every user and group and
confirm that the soft limit is not set greater than the hard limit.
091 WARNING: msgcnt x: mesg 091: V-2-91: vx_fcl_truncate - failure to

punch hole at offset offset for bytes bytes in File Change Log file; error
error_number
■ Description
The vxfs kernel has experienced an error while trying to manage
the space consumed by the File Change Log file. Because the space
cannot be actively managed at this time, the FCL has been
deactivated and has been truncated to 1 file system block, which
contains the FCL superblock.
■ Action
Re-activate the FCL.
092 WARNING: msgcnt x: mesg 092: V-2-92: vx_mkfcltran - failure to map

offset offset in File Change Log file
■ Description
The vxfs kernel was unable to map actual storage to the next offset
in the File Change Log file. This is mostly likely caused by a problem
with allocating to the FCL file. Because no new FCL records can be
written to the FCL file, the FCL has been deactivated.
■ Action
Re-activate the FCL.
Kernel messages
093 WARNING: msgcnt x: mesg 093: V-2-93: Disk layout versions older
than Version 6 are not supported for shared mounts. Upgrade to the
latest layout version.
See the Storage Foundation Release Notes.
■ Action
Upgrade your disk layout to Version 7 for shared mounts. Use the
vxupgrade command to begin upgrading file systems using older
disk layouts to Version 7.
096 WARNING: msgcnt x: mesg 096: V-2-96: file_system file system fullfsck
flag set - function_name.
■ Description
The next time the file system is mounted, a full fsck must be
performed.
■ Action
No immediate action required. When the file system is unmounted,
run a full file system check using fsck before mounting it again.
097 WARNING: msgcnt x: mesg 097: V-2-97: VxFS failed to create new
thread (error_number, function_address:argument_address)
■ Description
VxFS failed to create a kernel thread due to resource constraints,
which is often a memory shortage.
■ Action
VxFS will retry the thread creation until it succeeds; no immediate
action is required. Kernel resources, such as kernel memory, might
be overcommitted. If so, reconfigure the system accordingly.
098 WARNING: msgcnt x: mesg 098: V-2-98: VxFS failed to initialize File
Change Log for fileset fileset (index number) of mount_point file
system
■ Description
VxFS mount failed to initialize FCL structures for the current fileset
mount. As a result, FCL could not be turned on. The FCL file will
have no logging records.
■ Action
Reactivate the FCL.
Kernel messages
099 WARNING: msgcnt x: mesg 099: V-2-99: The specified value for
vx_ninode is less than the recommended minimum value of min_value
■ Description
Auto-tuning or the value specified by the system administrator
resulted in a value lower than the recommended minimum for the
total number of inodes that can be present in the inode cache. VxFS
will ignore the newly tuned value and will keep the value specified
in the message (VX_MINNINODE).
■ Action
Informational only; no action required.
101 WARNING: msgcnt x: mesg 101: V-2-101: File Change Log on

mount_point for file set index approaching max file size supported.
File Change Log will be reactivated when its size hits max file size
supported.
■ Description
The size of the FCL file is approching the maximum file size supported.
This size is platform specific. When the FCL file is reaches the
maximum file size, the FCL will be deactivated and reactivated. All
logging information gathered so far will be lost.
■ Action
Take any corrective action possible to restrict the loss due to the
FCL being deactivated and reactivated.
102 WARNING: msgcnt x: mesg 102: V-2-102: File Change Log of

mount_point for file set index has been reactivated.
■ Description
The size of FCL file reached the maximum supported file size and the
FCL has been reactivated. All records stored in the FCL file, starting
from the current fc_loff up to the maximum file size, have been purged.
New records will be recorded in the FCL file starting from offset
fs_bsize. The activation time in the FCL is reset to the time of
reactivation. The impact is equivalent to File Change Log being
deactivated and activated.
■ Action
Informational only; no action required.
About unique message identifiers
About unique message identifiers

VxFS generates diagnostic or error messages for issues not related to the kernel,
which are displayed along with a unique message identifier (UMI). Each message
has a description and a suggestion on how to handle or correct the underlying
problem. The UMI is used to identify the issue should you need to call Technical
Support for assistance.
Unique message identifiers

Some commonly encountered UMIs and the associated messages are described
on the following table:
Table B-2 Unique message identifiers and messages
20002 UX:vxfs command: ERROR: V-3-20002: message
■ Description
The command attempted to call stat() on a device path to ensure
that the path refers to a character device before opening the device,
but the stat() call failed. The error message will include the
platform-specific message for the particular error that was
encountered, such as "Access denied" or "No such file or directory".
■ Action
The corrective action depends on the particular error.
■ Description
The command attempted to open a disk device, but the open() call
failed. The error message includes the platform-specific message
for the particular error that was encountered, such as "Access
denied" or "No such file or directory".
■ Action
Table B-2 Unique message identifiers and messages (continued)
■ Description
The command attempted to read the superblock from a device, but
the read() call failed. The error message will include the
platform-specific message for the particular error that was
encountered, such as "Access denied" or "No such file or directory".
■ Action
■ Description
The command was invoked on a device that did not contain a valid
VxFS file system.
■ Action
Check that the path specified is what was intended.
■ Description
The command called stat() on a file, which is usually a file system
mount point, but the call failed.
■ Action
Check that the path specified is what was intended and that the
user has permission to access that path.
■ Description
The attempt to mount the file system failed because either the
request was to mount a particular Storage Checkpoint that does
not exist, or the file system is managed by an HSM and the HSM
is not running.
■ Action
In the first case, use the fsckptadm list command to see which
Storage Checkpoints exist and mount the appropriate Storage
Checkpoint. In the second case, make sure the HSM is running. If
the HSM is not running, start and mount the file system again.
■ Description
The attempt to mount a VxFS file system has failed because either
the volume being mounted or the directory which is to be the mount
point is busy.
The reason that a VxVM volume could be busy is if the volume is
in a shared disk group and the volume is currently being accessed
by a VxFS command, such as fsck, on a node in the cluster.
One reason that the mount point could be busy is if a process has
the directory open or has the directory as its current directory.
Another reason that the mount point could be busy is if the
directory is NFS-exported.
■ Action
For a busy mount point, if a process has the directory open or has
the directory as its current directory, use the fuser command to
locate the processes and either get them to release their references
to the directory or kill the processes. Afterward, attempt to mount
the file system again.
If the directory is NFS-exported, unexport the directory, such as
by using unshare mntpt on the Solaris operating system.
Afterward, attempt to mount the file system again.
■ Description
This message is printed by two different commands:
fsckpt_restore and mount. In both cases, the kernel's attempt
to mount the file system failed because of I/O errors or corruption
of the VxFS metadata.
■ Action
Check the console log for I/O errors and fix any problems reported
there. Run a full fsck.
■ Description
The mount options specified contain mutually-exclusive options,
or in the case of a remount, the new mount options differed from
the existing mount options in a way that is not allowed to change
in a remount.
■ Action
Change the requested mount options so that they are all mutually
compatible and retry the mount.
■ Description
Cluster mounts require the vxfsckd daemon to be running, which
is controlled by VCS.
■ Action
Check the VCS status to see why this service is not running. After
starting the daemon via VCS, try the mount again.
■ Description
In some releases of VxFS, before the VxFS mount command
attempts to mount a file system, mount tries to read the VxFS
superblock to determine the disk layout version of the file system
being mounted so that mount can check if that disk layout version
is supported by the installed release of VxFS. If the attempt to read
the superblock fails for any reason, this message is displayed. This
message will usually be preceded by another error message that
gives more information as to why the superblock could not be read.
■ Action
The corrective action depends on the preceding error, if any.
Appendix C
Disk layout
■ About disk layouts
■ About disk space allocation
■ VxFS Version 4 disk layout
■ Using UNIX Commands on File Systems Larger than One TB
About disk layouts

The disk layout is the way file system information is stored on disk. On VxFS, six
different disk layout versions were created to take advantage of evolving
technological developments.
The disk layout versions used on VxFS are:
Version 1 Version 1 disk layout is the original VxFS disk layout Not Supported
provided with pre-2.0 versions of VxFS.
Version 2 Version 2 disk layout supports features such as filesets, Not Supported
dynamic inode allocation, and enhanced security. The
Version 2 layout is available with and without quotas
support.
278 Disk layout
About disk space allocation
Version 3 Version 3 disk layout encompasses all file system Not Supported
structural information in files, rather than at fixed
locations on disk, allowing for greater scalability.
Version 3 supports files and file systems up to one
terabyte in size.
Version 4 Version 4 disk layout encompasses all file system Supported

structural information in files, rather than at fixed
locations on disk, allowing for greater scalability.
Version 4 supports files and file systems up to one
terabyte in size.
Version 5 Version 5 enables the creation of file system sizes up Supported

to 32 terabytes. Files can be a maximum of one terabyte.
File systems larger than 1TB must be created on a
Veritas Volume Manager volume.
Version 6 Version 6 disk layout enables features such as Supported

multi-volume support, cross-platform data sharing,
named data streams, and File Change Log.
Version 7 Version 7 disk layout enables support for variable and Supported
large size history log records.
Some of the disk layout versions were not supported on all UNIX operating systems.
Currently, only the Version 4, 5, 6, and 7 disk layouts can be created and mounted.
Version 1 and 2 file systems cannot be created nor mounted. Version 7 is the
default disk layout version.
The vxupgrade command is provided to upgrade an existing VxFS file system to
the Version 5, 6, or 7 layout while the file system remains online. You must upgrade
in steps from older to newer layouts.
The vxfsconvert command is provided to upgrade Version 1 and 2 disk layouts
to the Version 7 disk layout while the file system is not mounted.
See the vxfsconvert(1M) manual page.
About disk space allocation

Disk space is allocated by the system in 512-byte sectors. An integral number of
sectors are grouped together to form a logical block. VxFS supports logical block
sizes of 1024, 2048, 4096, and 8192 bytes. The default block size for file systems
less than one terabyte is 1024 bytes. The block size may be specified as an
argument to the mkfs utility and may vary between VxFS file systems mounted
Disk layout 279
VxFS Version 4 disk layout
on the same system. VxFS allocates disk space to files in extents. An extent is a
set of contiguous blocks.

Note: The information in this section also applies to the Version 5 disk layout.
The Version 4 disk layout allows the file system to scale easily to accommodate
large files and large file systems.
The original disk layouts divided up the file system space into allocation units.
The first AU started part way into the file system which caused potential alignment
problems depending on where the first AU started. Each allocation unit also had
its own summary, bitmaps, and data blocks. Because this AU structural information
was stored at the start of each AU, this also limited the maximum size of an extent
that could be allocated. By replacing the allocation unit model of previous versions,
the need for alignment of allocation units and the restriction on extent sizes was
removed.
The VxFS Version 4 disk layout divides the entire file system space into fixed size
allocation units. The first allocation unit starts at block zero and all allocation
units are a fixed length of 32K blocks. An exception may be the last AU, which
occupies whatever space remains at the end of the file system. Because the first
AU starts at block zero instead of part way through the file system as in previous
versions, there is no longer a need for explicit AU alignment or padding to be
added when creating a file system.
The Version 4 file system also moves away from the model of storing AU structural
data at the start of an AU and puts all structural information in files. So expanding
the file system structures simply requires extending the appropriate structural
files. This removes the extent size restriction imposed by the previous layouts.
All Version 4 structural files reside in the structural fileset.
The structural files in the Version 4 disk layout are:
object location Contains the object location table (OLT). The OLT, which is referenced
table file from the super-block, is used to locate the other structural files.
label file Encapsulates the super-block and super-block replicas. Although the
location of the primary super-block is known, the label file can be used
to locate super-block copies if there is structural damage to the file
system.
280 Disk layout
device file Records device information such as volume length and volume label,
and contains pointers to other structural files.
fileset header file Holds information on a per-fileset basis. This may include the inode
of the fileset's inode list file, the maximum number of inodes allowed,
an indication of whether the file system supports large files, and the
inode number of the quotas file if the fileset supports quotas. When
a file system is created, there are two filesets—the structural fileset
defines the file system structure, the primary fileset contains user
data.
inode list file Both the primary fileset and the structural fileset have their own set
of inodes stored in an inode list file. Only the inodes in the primary
fileset are visible to users. When the number of inodes is increased,
the kernel increases the size of the inode list file.
inode allocation Holds the free inode map, extended operations map, and a summary
unit file of inode resources.
log file Maps the block used by the file system intent log.
extent allocation Indicates the allocation state of each AU by defining whether each
unit state file AU is free, allocated as a whole (no bitmaps allocated), or expanded,
in which case the bitmaps associated with each AU determine which
extents are allocated.
extent allocation Contains the AU summary for each allocation unit, which contains
unit summary file the number of free extents of each size. The summary for an extent
is created only when an allocation unit is expanded for use.
free extent map Contains the free extent maps for each of the allocation units.
file
quotas files Contains quota information in records. Each record contains resources
allocated either per user or per group.
The Version 4 disk layout supports Access Control Lists and Block-Level
Incremental (BLI) Backup. BLI Backup is a backup method that stores and retrieves
only the data blocks changed since the previous backup, not entire files. This
saves times, storage space, and computing resources required to backup large
databases.
Figure C-1 shows how the kernel and utilities build information about the structure
of the file system.
The super-block location is in a known location from which the OLT can be located.
From the OLT, the initial extents of the structural inode list can be located along
with the inode number of the fileset header file. The initial inode list extents
Disk layout 281
contain the inode for the fileset header file from which the extents associated
with the fileset header file are obtained.
As an example, when mounting the file system, the kernel needs to access the
primary fileset in order to access its inode list, inode allocation unit, quotas file
and so on. The required information is obtained by accessing the fileset header
file from which the kernel can locate the appropriate entry in the file and access
the required information.
Figure C-1 VxFS Version 4 disk layout
Object Location Table

Fileset Header/ Initial Inode Extents
File Inode Number
Fileset Header
Initial Inode List File Inode
Extent Addresses
Inode List Inode
.... Inode Allocation

Unit Inode
OLT Replica ....
Primary Fileset Header Fileset Header File
Inode List inum Structural Fileset

Header
Fileset Index
and Name
Primary Fileset
max_inodes
Header
Features
....
....

VxFS disk layout Version 5 is similar to Version 4. Structural files in Version 5
are the same in Version 4. However, the Version 5 disk layout supports file systems
up to 32 terabytes. For a file system to take advantage of VxFS 32-terabyte support,
it must be created on a Veritas Volume Manager volume, and only on a 64-bit
282 Disk layout
kernel operating system. The maximum file system size on a 32-bit kernel is still
one terabyte. Files cannot exceed two terabytes in size. For 64-bit kernels, the
maximum size of the file system you can create depends on the block size:
If you specify the file system size when creating a file system, the block size
defaults to the appropriate value as shown above.
See the mkfs(1M) manual page.
The Version 5 disk layout also supports group quotas. Quota limits cannot exceed
one terabyte.
See “About quota files on &ProductNameLong;” on page 102.
Some UNIX commands may not work correctly on file systems larger than one
terabyte.
See “Using UNIX Commands on File Systems Larger than One TB” on page 283.

VxFS disk layout Version 6 is similar to Version 5. Structural files in Version 6
are the same in Version 5. The Version 6 disk layout can theoretically support
files and file systems up to 8 exabytes (26363). The maximum file system size that
can be created is currently restricted to 23535 blocks. For a file system to take
advantage of greater than 1 terabyte support, it must be created on a Veritas
Volume Manager volume. For 64-bit kernels, the maximum size of the file system
you can create depends on the block size:
Block Size Currently-Supported Maximum File System Size
1024 bytes 68,719,472,624 sectors (≈ 32 TB)
2048 bytes 137,438,945,248 sectors (≈ 64 TB)
4096 bytes 274,877,890,496 sectors (≈ 128 TB)
8192 bytes 549,755,780,992 sectors (≈ 256 TB)
The Version 6 disk layout also supports group quotas.

terabyte.
Disk layout 283

VxFS disk layout Version 7 is similar to Version 6. The Version 7 disk layout can
theoretically support files and file systems up to 8 exabytes (26363). The maximum
file system size that can be created is currently restricted to 23535 blocks. For a
file system to take advantage of greater than 1 terabyte support, it must be created
on a Veritas Volume Manager volume. For 64-bit kernels, the maximum size of
the file system you can create depends on the block size:
The Version 7 disk layout supports group quotas.
terabyte.
Using UNIX Commands on File Systems Larger than

One TB
terabyte.
The ustat command returns an EOVERFLOW error for VxFS files systems larger
than one terabyte because the variable used to store file system size overflows.
See the ustat(2) manual page.
System administration utilities such as backup may not operate correctly if they
are not large file aware (files larger than two gigabytes). Similarly, utilities that
operate at the file system level must be large file aware to operate correctly on
large file systems (file systems that are larger than one terabyte). Note also that
you can have a large file system without creating the file system with themkfs –o
largefiles option.
See the lfcompile(5) manual page.
284 Disk layout
Using UNIX Commands on File Systems Larger than One TB
Glossary
access control list (ACL) The information that identifies specific users or groups and their access privileges
for a particular file or directory.
agent A process that manages predefined Veritas Cluster Server (VCS) resource types.
Agents bring resources online, take resources offline, and monitor resources to
report any state changes to VCS. When an agent is started, it obtains configuration
information from VCS and periodically monitors the resources and updates VCS
with the resource status.
allocation unit A group of consecutive blocks on a file system that contain resource summaries,
free resource maps, and data blocks. Allocation units also contain copies of the
super-block.
API Application Programming Interface.
asynchronous writes A delayed write in which the data is written to a page in the system’s page cache,
but is not written to disk before the write returns to the caller. This improves
performance, but carries the risk of data loss if the system crashes before the data
is flushed to disk.
atomic operation An operation that either succeeds completely or fails and leaves everything as it
was before the operation was started. If the operation succeeds, all aspects of the
operation take effect at once and the intermediate states of change are invisible.
If any aspect of the operation fails, then the operation aborts without leaving
partial changes.
Block-Level Incremental A Symantec backup capability that does not store and retrieve entire files. Instead,
Backup (BLI Backup) only the data blocks that have changed since the previous backup are backed up.
buffered I/O During a read or write operation, data usually goes through an intermediate kernel
buffer before being copied between the user buffer and disk. If the same data is
repeatedly read or written, this kernel buffer acts as a cache, which can improve
performance. See unbuffered I/O and direct I/O.
contiguous file A file in which data blocks are physically adjacent on the underlying media.
data block A block that contains the actual data belonging to files and directories.
data synchronous A form of synchronous I/O that writes the file data to disk before the write returns,
writes but only marks the inode for later update. If the file size changes, the inode will
be written before the write returns. In this mode, the file data is guaranteed to be
286 Glossary
on the disk before the write returns, but the inode modification times may be lost
if the system crashes.
defragmentation The process of reorganizing data on disk by making file data blocks physically
adjacent to reduce access times.
direct extent An extent that is referenced directly by an inode.
direct I/O An unbuffered form of I/O that bypasses the kernel’s buffering of data. With direct
I/O, the file system transfers data directly between the disk and the user-supplied
buffer. See buffered I/O and unbuffered I/O.
discovered direct I/O Discovered Direct I/O behavior is similar to direct I/O and has the same alignment
constraints, except writes that allocate storage or extend the file size do not require
writing the inode changes before returning to the application.
encapsulation A process that converts existing partitions on a specified disk to volumes. If any
partitions contain file systems, /etc/filesystems entries are modified so that the
file systems are mounted on volumes instead. Encapsulation is not applicable on
some systems.
extent A group of contiguous file system data blocks treated as a single unit. An extent
is defined by the address of the starting block and a length.
extent attribute A policy that determines how a file allocates extents.
external quotas file A quotas file (named quotas) must exist in the root directory of a file system for
quota-related commands to work. See quotas file and internal quotas file.
file system block The fundamental minimum size of allocation in a file system. This is equivalent
to the fragment size on some UNIX file systems.
fileset A collection of files within a file system.
fixed extent size An extent attribute used to override the default allocation policy of the file system
and set all allocations for a file to a specific fixed size.
fragmentation The on-going process on an active file system in which the file system is spread
further and further along the disk, leaving unused gaps or fragments between
areas that are in use. This leads to degraded performance because the file system
has fewer options when assigning a file to an extent.
GB Gigabyte (230 bytes or 1024 megabytes).
hard limit The hard limit is an absolute limit on system resources for individual users for
file and data block usage on a file system. See quota.
indirect address extent An extent that contains references to other extents, as opposed to file data itself.
A single indirect address extent references indirect data extents. A double indirect
address extent references single indirect address extents.
indirect data extent An extent that contains file data and is referenced via an indirect address extent.
Glossary 287
inode A unique identifier for each file within a file system that contains the data and
metadata associated with that file.
inode allocation unit A group of consecutive blocks containing inode allocation information for a given
fileset. This information is in the form of a resource summary and a free inode
map.
intent logging A method of recording pending changes to the file system structure. These changes
are recorded in a circular intent log file.
internal quotas file VxFS maintains an internal quotas file for its internal usage. The internal quotas
file maintains counts of blocks and indices used by each user. See quotas and
external quotas file.
K Kilobyte (210 bytes or 1024 bytes).
large file A file larger than two one terabyte. VxFS supports files up to 8 exabytes in size.
large file system A file system larger than one terabytes. VxFS supports file systems up to 8 exabytes
in size.
latency For file systems, this typically refers to the amount of time it takes a given file
system operation to return to the user.
metadata Structural data describing the attributes of files on a disk.
MB Megabyte (220 bytes or 1024 kilobytes).
mirror A duplicate copy of a volume and the data therein (in the form of an ordered
collection of subdisks). Each mirror is one copy of the volume with which the
mirror is associated.
multi-volume file A single file system that has been created over multiple volumes, with each volume
system having its own properties.
MVS Multi-volume support.
object location table The information needed to locate important file system structural elements. The
(OLT) OLT is written to a fixed location on the underlying media (or disk).
object location table A copy of the OLT in case of data corruption. The OLT replica is written to a fixed
replica location on the underlying media (or disk).
page file A fixed-size block of virtual address space that can be mapped onto any of the
physical addresses available on a system.
preallocation A method of allowing an application to guarantee that a specified amount of space
is available for a file, even if the file system is otherwise out of space.
primary fileset The files that are visible and accessible to the user.
quotas Quota limits on system resources for individual users for file and data block usage
on a file system. See hard limit and soft limit.
288 Glossary
quotas file The quotas commands read and write the external quotas file to get or change
usage limits. When quotas are turned on, the quota limits are copied from the
external quotas file to the internal quotas file. See quotas, internal quotas file,
and external quotas file.
reservation An extent attribute used to preallocate space for a file.
root disk group A special private disk group that always exists on the system. The root disk group
is named rootdg.
shared disk group A disk group in which the disks are shared by multiple hosts (also referred to as
a cluster-shareable disk group).
shared volume A volume that belongs to a shared disk group and is open on more than one node
at the same time.
snapshot file system An exact copy of a mounted file system at a specific point in time. Used to do
online backups.
snapped file system A file system whose exact image has been used to create a snapshot file system.
soft limit The soft limit is lower than a hard limit. The soft limit can be exceeded for a limited
time. There are separate time limits for files and blocks. See hard limit and quotas.
Storage Checkpoint A facility that provides a consistent and stable view of a file system or database
image and keeps track of modified data blocks since the last Storage Checkpoint.
structural fileset The files that define the structure of the file system. These files are not visible or
accessible to the user.
super-block A block containing critical information about the file system such as the file
system type, layout, and size. The VxFS super-block is always located 8192 bytes
from the beginning of the file system and is 8192 bytes long.
synchronous writes A form of synchronous I/O that writes the file data to disk, updates the inode
times, and writes the updated inode to disk. When the write returns to the caller,
both the data and the inode have been written to disk.
TB Terabyte (240 bytes or 1024 gigabytes).
transaction Updates to the file system structure that are grouped together to ensure they are
all completed.
throughput For file systems, this typically refers to the number of I/O operations in a given
unit of time.
unbuffered I/O I/O that bypasses the kernel cache to increase I/O performance. This is similar to
direct I/O, except when a file is extended; for direct I/O, the inode is written to
disk synchronously, for unbuffered I/O, the inode update is delayed. See buffered
I/O and direct I/O.
Glossary 289
volume A virtual disk which represents an addressable range of disk blocks used by
applications such as file systems or databases.
volume set A container for multiple different volumes. Each volume can have its own
geometry.
vxfs The Veritas File System type. Used as a parameter in some commands.
VxFS Veritas File System.
VxVM Veritas Volume Manager.
290 Glossary
Index
A cpio_vxfs 58
access control lists 20 creating a multi-volume support file system 122
alias for Quick I/O files 187 creating file systems with large files 38
allocation policies 56 creating files with mkfs 211–212
default 56 creating Quick I/O files 188
extent 15 cron 26, 42
extent based 14 cron sample script 43
multi-volume support 125
D
B data copy 62
bad block revectoring 33 data integrity 18
blkclear 18 data Storage Checkpoints definition 72
blkclear mount option 33 data synchronous I/O 34, 63
block based architecture 23 data transfer 62
block size 14, 279 default
blockmap for a snapshot file system 99 allocation policy 56
buffered file systems 18 block sizes 14, 279
buffered I/O 63 default_indir_size tunable parameter 46
defragmentation 26
extent 42
C scheduling with cron 42
cache advisories 64 delaylog mount option 31–32
Cached Quick I/O 197 device file 280
Cached Quick I/O read-ahead 197 direct data transfer 62
cio direct I/O 62
Concurent I/O 39 directory reorganization 43
closesync 18 disabled file system
cluster mount 22 snapshot 100
commands transactions 229
cron 26 discovered direct I/O 63
fsadm 26 discovered_direct_iosize tunable parameter 46
getext 58 disk layout
mkfs 279 Version 1 277
qiostat 199 Version 2 277
setext 58 Version 3 278
contiguous reservation 57 Version 4 278–279
converting a data Storage Checkpoint to a nodata Version 5 278, 281
Storage Checkpoint 78 Version 6 278
convosync mount option 31, 35 Version 7 278
copy-on-write technique 67, 71 disk space allocation 14, 279
cp_vxfs 58 displaying mounted file systems 216
292 Index
Dynamic Storage Tiering file (continued)

multi-volume support 119 sparse 57
file change log 47
E file system
block size 59
enabling Quick I/O 196
buffering 18
encapsulating volumes 119
displaying mounted 216
enhanced data integrity modes 18
increasing size 220
ENOENT 233
fileset
ENOSPC
header file 280
86
primary 69
ENOTDIR 233
filesystems file 215
expansion 26
fixed extent size 55
extensions of Quick I/O files 187
fixed write size 57
extent 14, 55
fragmentation
attributes 55
monitoring 42–43
description 279
reorganization facilities 42
indirect 15
reporting 42
reorganization 43
fragmented file system characteristics 43
extent allocation 14–15
free extent map file 280
aligned 56
free space monitoring 42
control 55
freeze 65
fixed size 55
freezing and thawing, relation to Storage
unit state file 280
Checkpoints 69
unit summary file 280
fsadm 26
extent size
how to reorganize a file system 222
indirect 15
how to resize a file system 220
external quotas file 102
reporting extent fragmentation 43
scheduling defragmentation using cron 43
F fsadm_vxfs 39
fc_foff 111 fscat 94
fcl_inode_aging_count tunable parameter fsck 78
49 fsckptadm
fcl_inode_aging_size tunable parameter 49 Storage Checkpoint administration 74
fcl_keeptime tunable parameter 47 fstab file
fcl_maxalloc tunable parameter 47 editing 215
fcl_winterval tunable parameter 48 fstyp
file how to determine the file system type 217–218
device 280 fsvoladm 122
extent allocation unit state 280
extent allocation unit summary 280
fileset header 280
G
get I/O parameter ioctl 65–66
free extent map 280
getext 58
inode allocation unit 280
getfacl 20
inode list 280
global message IDs 230
intent log 280
label 279
object location table 279 H
quotas 280 how to access a Storage Checkpoint 77
Index 293
how to create a backup file system 223 K

how to create a Storage Checkpoint 75 kernel tunable parameters 40
how to determine the file system type 217–218
how to display mounted file systems 216
how to edit the fstab file 215
L
how to edit the vfstab file 215 label file 279
how to mount a Storage Checkpoint 77 large files 20, 37
how to remove a Storage Checkpoint 76 creating file systems with 38
how to reorganize a file system 222 mounting file systems with 38
how to resize a file system 220 largefiles mount option 38
how to restore a file system 225 local mount 22
how to set up user quotas 226 log failure 230
how to turn off quotas 227 log mount option 30
how to turn on quotas 226 logiosize mount option 33
how to unmount a Storage Checkpoint 78
how to view quotas 227 M
HSM agent error message 256–257 max_direct_iosize tunable parameter 50
hsm_write_prealloc 48 max_diskq tunable parameter 50
max_seqio_extent_size tunable parameter 50
I maximum I/O size 41
I/O metadata
direct 62 multi-volume support 119
sequential 63 mincache mount option 31, 34
synchronous 63 mkfs 279
I/O requests creating files with 211–212
asynchronous 34 creating large files 39
synchronous 33 modes
increasing file system size 220 enhanced data integrity 18
indirect extent monitoring fragmentation 42
address size 15 mount 18, 39
double 15 how to display mounted file systems 216
single 15 how to mount a file system 213
initial_extent_size tunable parameter 49 mounting a Storage Checkpoint 77
inode allocation unit file 280 pseudo device 77
inode list error 230 mount options 30
inode list file 280 blkclear 33
inode table 40 choosing 30
internal 40 combining 39
sizes 40 convosync 31, 35
inodes, block based 15 delaylog 19, 31–32
intent log 16 extended 17
file 280 largefiles 38
multi-volume support 119 log 18, 30
Intent Log Resizing 17 logiosize 33
internal inode table 40 mincache 31, 34
internal quotas file 102 nodatainlog 31, 33
ioctl interface 55 tmplog 32
mounted file system
displaying 216
294 Index
mounting a file system 213 Quick I/O (continued)

option combinations 39 restrictions 188
with large files 38 special naming convention 187
msgcnt field 231 Quick I/O files
multi-volume support 118 access regular UNIX files 190
creating a MVS file system 122 preallocating space 191
multiple block operations 15 statistics 199
mv_vxfs 58 using relative and absolute path names 190
quota commands 103
N quotacheck 104
quotas 101
name space
exceeding the soft limit 102
preserved by Storage Checkpoints 68
hard limit 101
naming convention, Quick I/O 187
92
ncheck 114
soft limit 101
nodata Storage Checkpoints 78
quotas file 102, 280
nodata Storage Checkpoints definition 73
quotas.grp file 102
nodatainlog mount option 31, 33
O R
read-ahead functionality in Cached Quick I/O 197
O_SYNC 31
read-only Storage Checkpoints 77
object location table file 279
read_ahead 51
read_nstream tunable parameter 45
P read_pref_io tunable parameter 45
parameters relative and absolute path names used with symbolic
default 44 links 190
tunable 45 removable Storage Checkpoints definition 74
tuning 44 reorganization
performance directory 43
overall 30 extent 43
snapshot file systems 97 report extent fragmentation 42
preallocating space for Quick I/O files 191 reservation space 55
primary fileset relation to Storage Checkpoints 69 restrictions on Quick I/O 188
pseudo device 77 Reverse Path Name Lookup 114
Q S
qio module sectors
loading on system reboot 200 forming logical blocks 279
qio_cache_enable tunable parameter 50, 197 sequential I/O 63
qiomkfile 188 setext 58
qiostat 199 setfacl 20
Quick I/O 185 snapped file systems 20, 93
access Quick I/O files as raw devices 187 performance 97
access regular UNIX files 190 unmounting 94
creating Quick I/O files 188 snapread 94
direct I/O 186 snapshot 223
double buffering 187 how to create a backup file system 223
extension 187
read/write locks 186
Index 295
snapshot file system SVID requirement

on CFS 94 VxFS conformance to 27
snapshot file systems 20, 93 symbolic links
blockmap 99 accessing Quick I/O files 190
data block area 99 synchronous I/O 63
disabled 100 system failure recovery 16
errors 243 system performance
fscat 94 overall 30
fuser 94
multiple 94 T
performance 97
temporary directories 19
read 94
thaw 65
super-block 99
tmplog mount option 32
snapsize 95
transaction disabling 229
sparse file 57
tunable I/O parameters 45
statistics
default_indir_size 46
generated for Quick I/O 199
discovered_direct_iosize 46
storage
fcl_keeptime 47
clearing 33
fcl_maxalloc 47
uninitialized 33
fcl_winterval 48
Storage Checkpoints
initial_extent_size 49
accessing 77
inode_aging_count 49
administration of 74
inode_aging_size 49
converting a data Storage Checkpoint to a nodata
max_direct_iosize 50
Storage Checkpoint with multiple Storage
max_diskq 50
Checkpoints 82
max_seqio_extent_size 50
creating 75
qio_cache_enable 50, 197
data Storage Checkpoints 72
read_nstream 45
definition of 67
read_pref_io 45
difference between a data Storage Checkpoint
Volume Manager maximum I/O size 41
and a nodata Storage Checkpoint 79
write_nstream 46
freezing and thawing a file system 69
write_pref_io 45
mounting 77
write_throttle 52
multi-volume support 119
tuning I/O parameters 44
nodata Storage Checkpoints 73, 78
typed extents 15
operation failures
86
pseudo device 77 U
read-only Storage Checkpoints 77 umount command 216
removable Storage Checkpoints 74 uninitialized storage, clearing 33
removing 76 unmount 78, 230
space management a snapped file system 94
86
synchronous vs. asynchronous conversion 79 V
types of 72 VEA 25
unmounting 78 VERITAS Enterprise Administrator 25
using the fsck command 78 Version 1 disk layout 277
writable Storage Checkpoints 77 Version 2 disk layout 277
super-block 99 Version 3 disk layout 278
296 Index
Version 4 disk layout 278–279

Version 5 disk layout 278, 281
Version 6 disk layout 278
Version 7 disk layout 278
vfstab file
editing 215
virtual disks 27
vol_maxio tunable I/O parameter 41
volume sets 120
VOP_INACTIVE 246
VX_DSYNC 64
VX_FREEZE 65, 104
VX_FULLFSCK 230, 232–236, 240–241, 243, 246–
247, 249–250, 253–256, 265
VX_GETCACHE 64
VX_SETCACHE 64
VX_SNAPREAD 94
VX_THAW 65
VX_UNBUFFERED 63
vxdump 58
vxedquota
how to set up user quotas 226
VxFS
storage allocation 29
vxfs_inotopath 114
vxfs_ninode 40
vxfsu_fcl_sync 48
vxlsino 114
vxquota
how to view quotas 227
vxquotaoff
how to turn off quotas 227
vxquotaon 226
vxrestore 58, 225
vxtunefs
changing extent size 15
vxvset 120
W
writable Storage Checkpoints 77
write size 57
write_nstream tunable parameter 46
write_pref_io tunable parameter 45
write_throttle tunable parameter 52

VXFS 5.0

Uploaded by

Document Informationclick to expand document information

Document Informationclick to expand document information

Copyright:

Available Formats

VXFS 5.0

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

VXFS 5.0

Uploaded by

Copyright:

Available Formats

Veritas File System

Veritas File System 5.0

All rights reserved.

Federal acquisitions: Commercial Software - Government Users Subject to Standard License

Symantec Corporation 20330 Stevens Creek Blvd. Cupertino, CA 95014 USA

Printed in the United States of America.

Chapter 1 Introducing Veritas File System

Chapter 2 VxFS performance: creating, mounting, and tuning

Chapter 3 Extent attributes

Chapter 4 VxFS I/O Overview

Unbuffered I/O ...................................................................... 63

Chapter 5 Storage Checkpoints

Chapter 6 Online backup using file system snapshots

Chapter 8 File Change Log

Chapter 9 Multi-volume file systems

Volume encapsulation ................................................................. 129

Chapter 10 Dynamic Storage Tiering

File placement policies and extending files ...................................... 184

Chapter 11 Quick I/O for Databases

Appendix A Quick Reference

Identifying file system types ......................................................... 217

Appendix B Diagnostic messages

Appendix C Disk layout

■ About Veritas File System

■ Veritas File System features

■ Veritas File System performance enhancements

■ Using Veritas File System

About Veritas File System

■ File system disk layouts

File system disk layouts

Veritas File System features

■ Cluster File System

starting block Is the starting file system block of the extent.

number of blocks Is the number of contiguous blocks in the extent.

Typed extents have the following characteristics:

Fast file system recovery

VxFS intent log

VxFS intent log resizing

See the mkfs_vxfs(1M) and the fsadm_vxfs(1M) manual pages.

Extended mount options

■ Large file sizes

Enhanced data integrity modes

The blkclear option and data integrity

The closesync option and data integrity

The log option and data integrity

Enhanced performance mode

The delaylog option and enhanced performance

Modes of temporary file systems

Improved synchronous writes

Warning: The use of the -o convosync=dsync option violates POSIX semantics.

Support for large files

Access Control Lists

■ Backing up the entire file system (using fscat)

Support for databases

Cluster file systems

Cross-platform data sharing

File Change Log