An optional set of utilities is available for AdvFS that expands the capabilities of the file system. The POLYCENTER Advanced File System Utilities (AdvFS Utilities) provide functions such as adding volumes without reconfiguring the directory hierarchy of the file system, cloning filesets to enable online backups, and improving system performance with file defragmentation, domain balancing, and file striping. The AdvFS Utilities also include a graphical user interface (GUI) that simplifies file system management.
The AdvFS component is licensed with the Digital UNIX operating system and is available as an optional subset during an advanced installation. You can choose AdvFS as the file system type for the root, /usr, or /var file systems. See the Installation Guide for more information about performing advanced installations. The AdvFS Utilities are available as a separately licensed layered product.
Before setting up AdvFS, you need to understand how it differs from traditional UNIX file systems. These differences, although minor with regard to your transition from a UNIX File System (UFS), play a role in how you plan and maintain AdvFS. For information on UFS, see Chapter 7.
The remaining sections in this chapter describe the unique characteristics of the file system design, instructions on setting up a new file system, and a clarification of what you can accomplish without the optional file system utilities.
Table 8-1 lists and describes the commands unique to the base portion of AdvFS.
AdvFS uses the the standard UFS quota commands to establish and manage AdvFS user and group quotas. Table 8-2 lists and describes the user and group quota commands to use with AdvFS.
Table 8-3 lists and describes the optional Advanced File System Utilities that are available as a separately licensed layered product.
The new design overcomes many unexpected events, such as a sudden system
interruption or running out of disk space.
System administrators can manage disk capacity independently of the
logical file system directory structure and transparently to the user. As
a result, configuration planning is less complicated and more flexible. A
system administrator can add or remove disks while the file system is active.
End users can retrieve their own unintentionally deleted files from
predefined trashcan directories.
Logical structures, quota controls, and backup capabilities are familiar,
which promotes a smooth transition to the new file system.
From a user's perspective, AdvFS behaves like any UNIX file system.
End users can use the mkdir command to create new directories,
the cd command to change directories, and the ls command
to list directory contents. Application programmers encounter a programming
interface based on UNIX: open(), close(), creat(), and so on. AdvFS is POSIX 1003.1 compliant.
AdvFS replaces or eliminates several standard commands, such as newfs, dump, restore, and fsck.
Rebooting after a system interruption is fast. AdvFS uses write-ahead
logging, instead of the fsck utility, as a way to check for and
repair file system inconsistencies that can occur during an unexpected system
crash or power failure. The number of uncommitted records in the log, not
the amount of data in the file system, dictates the speed of a recovery.
As a result, reboots are quick and predictable, measured in seconds instead
of minutes. User file data can be logged as well, providing better control
over data recovery.
To maintain data availability, system administrators can perform backups,
file system reconfiguration, and file system tuning without taking the system
off line.
An extent-based file allocation scheme allows for fewer and larger I/O
transfers, which increases sequential read/write throughput and simplifies
large data transfers. The file system performs large reads from disk when
sequential access is anticipated. It also performs large writes by combining
adjacent data into a single disk transfer.
Table 8-4 lists AdvFS features and their benefits,
which are unavailable to traditional file systems such as the BSD or the System
V file systems. Note that some of the features listed require the optional
POLYCENTER AdvFS Utilities license.
8.1 Features and Benefits
The AdvFS and AdvFS Utilities
provide an innovative design that is not based on any existing file system,
such as the BSD or the System V file systems. AdvFS enables you to:
The AdvFS Utilities enable multivolume file systems. Multiple
volumes allow file-level striping and file migration. File-level striping
improves file transfer rates by spreading I/O among several disks. File migration
allows for load and capacity balancing and reduces file fragmentation.
The decoupled file system structure enables you to manage the physical
storage layer apart from the directory hierarchy layer. This means that you
can move files between a defined group of disks, without changing file pathnames.
Since the pathnames remain the same, the action is completely transparent
to end users.
The two-layer structure is the cornerstone of AdvFS. Supporting the
design are two file system concepts: the file domain and the fileset.
Creating a file domain is the first step in setting up AdvFS. The number
of file domains that you construct on your system depends on the needs of
your site, the resources available to the system, and the number of independent
file systems you choose to manage.
When created, all file domains consist of a single (or initial) volume.
You transform a single-volume file domain into a multiple-volume file domain
by adding one or more volumes. Unless you have installed the optional file
system utilities and registered the license PAK, you are limited to creating
single-volume file domains.
The /etc/fdmns directory, which the file system automatically
creates and updates for you, contains a subdirectory for each file domain
on your system. The subdirectories contain a symbolic link to every volume
in the file domain.
Use the following guidelines for file domains:
To reduce the risk of domain failure, limit the number of
volumes to 3 volumes per file domain.
Filesets and file systems are equivalent in many ways. For instance,
you mount filesets individually like you mount file systems. Similarly, filesets
are units on which you enable quotas and back up data.
Although you can set up a fileset to simulate a traditional file system,
filesets offer more flexibility. For instance, you can create multiple filesets
in place of one file system, allowing you to manage each fileset independently.
Conversely, you can create one large fileset in a situation where you usually
define multiple file systems, thereby reducing management overhead.
Unlike file systems, filesets can have clone filesets. A clone fileset
is a read-only copy of an existing fileset that you create to capture your
data at one moment in time. You can back up the contents of the clone fileset
to media while the original fileset remains available to system users. The
clone fileset utility, clonefset, is available with the POLYCENTER
Advanced File System Utilities.
Use the following guidelines for filesets:
For multivolume file domains, new files are allocated sequentially across
volumes. Volumes that are more than 86% full (allocated) are not used for
new file allocation unless all volumes are over 86% full. When existing files
are appended, storage is allocated on the volume on which the file was initially
allocated, until that volume is full.
When a new volume is added to a file domain, it is added to the storage
allocation sequence. Files are allocated to the new volume in turn.
When a disk is fragmented, AdvFS writes data to isolated physical pages,
based on availability, instead of writing to contiguous pages.
Many users on a system increases file activity. As a result, files
become more fragmented on the disk.
File fragmentation can reduce the I/O performance of AdvFS. You can
use the defragment utility to reduce domain fragmentation. The defragment utility is available with the optional AdvFS Utilities. See
the
As you begin planning, decide whether you want to set up AdvFS to resemble
a traditional UFS configuration. Once you become familiar with AdvFS, you
can begin to move away from the traditional model.
When planning your configuration, consider setting up the root and /usr file systems on AdvFS. Using AdvFS as the root file system enables
booting from an AdvFS file domain and makes AdvFS features available on all
local file systems. By having the /usr file system on AdvFS you
can significantly reduce the amount of time your system is down after a system
failure.
You can put the root file system and /usr file system on
AdvFS during the initial base system installation. If you prefer, you can
convert your existing root and /usr file systems after installation.
See Section 8.9 and Section 8.10
for conversion guidelines.
The following procedure is a guide for setting up an active, single-volume
file system:
You can add more volumes to any existing file domain (except for the root_domain) by using the addvol utility, if you have installed
the file system utilities.
Name each fileset the same as its mount-point directory; for example,
if the mount-point directory is /tmp, name the fileset tmp.
In the unlikely event of a severe failure where you must restore the /fdmns directory manually by reconstructing that directory, you must
have a separate record of your file system configuration with the name of
each file domain and its associated volumes. Always keep this record up-to-date.
The following examples use AdvFS to
set up active file systems. The file-domain configurations in the examples
are:
This example mounts one fileset, called usr, on the /usr mount-point directory, which already exists on the system:
This example creates a single volume domain, domain2, and
two filesets in the domain, tmp and public. Because
the domain has only one volume, the files in both filesets reside on one volume.
This is an accepted configuration on AdvFS.
To set up this configuration, the example mounts two filesets, called tmp and public, on the respective mount-point directories:
This example mounts one fileset, called projects, on the /projects mount-point directory, which the example creates:
Refer to the
The AdvFS quota system is compatible with the Berkeley-style quotas
of UFS. Basically, AdvFS supports user account and group quotas. However,
the AdvFS quota system differs in two ways: AdvFS differentiates between quota
maintenance and quota enforcement and supports fileset quotas.
AdvFS quota maintenance tracks file and disk space usage. It keeps
a record of the number of files and blocks a user or group is using in the quota.user and quota.group files found in the root directory
of a fileset. The AdvFS quota system always maintains quota information.
Unlike UFS, this function cannot be disabled.
When quota enforcement is enabled, the AdvFS quota system enforces all
quota limits set by the system administrator. You use the edquota
command to set quota limits. Use the quotaon and quotaoff commands to enable and disable quota enforcement.
The AdvFS user and group quota commands are the same as UFS quota commands. Table 8-5 lists the quota commands used to set and maintain
user and group quotas.
In addition to the Berkeley-style user and group quotas, AdvFS also
supports fileset quotas. Fileset quotas are similar to user/group quotas
in both function and management.
Fileset quotas apply to the fileset rather than individual users and/or
groups. You can use them to limit the amount of disk storage and number of
files consumed by a fileset. This is useful when a file domain contains several
filesets. Without fileset quotas, all filesets have access to all disk space
in a file domain, allowing one fileset to use all the disk space in a file
domain.
Filesets can have both soft and hard disk storage and file limits.
When a hard limit is reached, no more disk space allocations or file creations
that would exceed the limit are allowed. The soft limit may be exceeded for
a period of time (called the grace period). If the soft limit is exceeded
for an amount of time that exceeds the grace period, no more disk space allocations
or file creations are allowed until enough disk space is freed or enough files
are deleted to bring the disk space usage or number of files below the soft
limit. The grace periods for the soft limits are set with the edquota
-tg command (the -g switch must be used because AdvFS uses
the group quota file of the fileset to maintain fileset quotas). Fileset quotas
are managed using the commands shown in Table 8-6.
If you already use the dump or restore commands
to back up and restore data, then the vdump and vrestore
commands will be familiar to you.
There are differences between the UFS commands and AdvFS commands.
The vdump command supports other file system types, so you can
use vdump command capabilities on files systems other than AdvFS.
Several minor flags are absent from AdvFS commands; new flags increase your
access to information when backing up and restoring files. The vdump command also provides the following features that are unavailable with
the dump command:
This feature expands the support of the vdump command to
other file systems, including UFS. This means that you can simplify the task
of backing up multiple file systems by using the same backup facility across
the system. Other file systems can benefit from the extended features provided
by the vdump command.
This feature refines the granularity of your system backups. Instead
of backing up an entire file system, you can selectively back up individual
subdirectories by using the -D flag.
This feature writes data in a compressed form to a saveset, which reduces
storage usage and runs faster on slow backup devices by writing less data.
Table 8-7 lists and describes new command flags for
the vdump command.
8.2 AdvFS Design Overview
Unlike UFS, AdvFS consists of two
distinct layers; each layer contains different file system mechanisms. The
directory hierarchy layer implements the file-naming scheme and Digital UNIX
file system-compliant functions such as creating and opening files or reading
and writing to files. The physical storage layer implements write-ahead logging,
caching, file allocation, and physical disk I/O functions.
8.2.1 File Domains
The following sections describe file domains and filesets in more detail.
A file domain is a named set of one
or more volumes that provides a shared pool of physical storage. With respect
to file domains, a volume is any mechanism that behaves like a UNIX block
device: an entire magnetic disk, a disk partition, or a logical volume that
is configured with the Logical Volume Manager (LVM) or the Logical Storage
Manager (LSM).
Refer to the addvol(8), mkfdmn(8), advfs(4), fdmns(4),
and showdmn(8) reference pages for more information.
8.2.2 Filesets and File Systems
A fileset represents a portion of the directory
hierarchy of a file system; each fileset, which has a unique name, is a collection
of directories and files that form a subtree structure. Because the hierarchy
layer is independent of the storage layer, you can manage file placement without
affecting the logical structure of filesets.
Refer to the advfs(4), fdmns(4), mkfset(8), and showfsets(8)
reference pages for more information.
8.3 File Storage Allocation
The Advanced File
System (AdvFS) always attempts to write each file to disk as a set of contiguous
units called pages; a page is 8 KB of disk space. Contiguous, in this context,
means storage on disk that is physically adjacent. A set of one or more contiguous
pages is called an extent. Contiguous placement of the pages means that the
I/O mechanism works more efficiently. When a file consists of many small
extents, the I/O mechanism must work harder to read or write that file.
8.3.1 Allocation Policy
Files are not static; disk space requirements change over time. To
maintain contiguous file placement without over-allocating space on the disk,
AdvFS applies a policy to file storage allocation. Each time a file is appended,
AdvFS adds pages to the file by preallocating one-fourth of the file size
up to 16 pages. If a large write requires more file space, AdvFS attempts
to allocate up to 256 contiguous pages. Excess preallocated space is truncated
when the file is closed. Unused preallocated space is then available for
the next write.
8.3.2 Fragments
AdvFS writes files to disk in sets of 8-KB pages. When a file uses
only part of the last page, less than 8 KB, a file fragment is
created. The fragment, which is from 1 KB to 7 KB in size, is allocated from
the frag file. Using fragments considerably reduces the amount
of unused, wasted disk space. Note that the frag file is a special file not
visible in the directory hierarchy.
8.3.3 Policy Allocation Limitations
Given the dynamic nature of a file system, the file storage-allocation
policy cannot always guarantee contiguous page placement. The following factors
affect the policy:
defragment(8) reference page for information on reducing file fragmentation.
8.4 Setting Up the Advanced File System
The number sign (#) between the
file domain and fileset is part of the syntax that represents a fileset; this
character does not indicate a comment.
# mkfdmn /dev/rz3c domain1
# mkfset domain1 usr
# mount -t advfs domain1#usr /usr
# mkfdmn /dev/rz2c domain2
# mkfset domain2 tmp
# mkfset domain2 public
# mkdir /public
# mount -t advfs domain2#tmp /tmp
# mount -t advfs domain2#public /public
# mkfdmn /dev/rz1c domain3
# mkfset domain3 projects
# mkdir /projects
# mount -t advfs domain3#projects /projects
mkfdmn(8), mkfset(8), and addvol(8) reference pages
for more information.
8.5 Managing File System and Fileset Quotas
AdvFS eliminates the slow reboot
activities associated with quotas on UFS. As a result, enabling quotas is
a useful way of tracking and controlling the amount of physical storage that
each fileset consumes.chfsets(8,) showfsets(8,) showfdmn(8,) edquota(8), ncheck(8), quot(8), quota(1), quotacheck(8), quotaon(8),
and vrepquota(8) reference pages for more information.
8.6 Backing Up Data
The dump command supports
UFS exclusively. As a result, AdvFS provides an equivalent backup command
called vdump. AdvFS also replaces the restore command
with the equivalent vrestore command.
vdump(8) and vrestore(8)
reference pages for more information.
AdvFS creates a corresponding subdirectory each time you create a file
domain. For example, you can create a file domain called mydomain,
which contains the volume /dev/rz1c. The file system creates the /etc/fdmns/mydomain subdirectory, which contains a symbolic link to /dev/rz1c. When you add or remove a volume from the file domain, the
file system updates the subdirectory by adding or removing symbolic links.
In some ways, the /etc/fdmns directory resembles the /etc/fstab file; each has special significance and requires extra attention.
You must restore the /etc/fdmns directory if its contents are
deleted, corrupted, or if you install a new version of the operating system.
Although a missing or damaged /etc/fdmns directory prevents access
to the file domain, the data within the file domain remains intact.
Always back up the /etc/fdmns directory whenever you create
a new file domain, add a volume to an existing file domain, or remove a volume
from an existing file domain.
If you choose to reconstruct manually, you must know the name of each
file domain on your system and its associated volumes. In other words, you
need detailed records of the file domains on your system.
The following example manually reconstructs two file domains, each containing
a single volume (or special device). The file domains are:
The following example, which requires that the optional AdvFS Utilities
be installed, reconstructs one multivolume file domain. The domain1
file domain contains three volumes:
You can use the advscan command to rebuild all or a part
of your /etc/fdmns file domain. The advscan command
can perform the following tasks:
AdvFS uses write-ahead logging as a way to reduce the impact of system
interruptions. As your system reboots, the file system scans all records
in the recovery log. Any operations that were uncommitted when the interruption
occurred are undone. Thus, the number of uncommitted records in the log determines
the speed of the recovery. Since the recovery depends on the number of records
in the log, instead of the amount of data in the file system, the recovery
process dramatically improves. The default log size is 4 MB.
AdvFS automatically initiates crash recovery on a file domain as soon
as you mount a fileset within that file domain. You can add filesets to the /etc/fstab file (at least one fileset per domain) so that all file domains
recover during the system reboot.
Refer to the
This section presents instructions for converting the root file system
from UFS to AdvFS. These instructions are guidelines, that is, suggestions
to illustrate the process of converting the root file system to AdvFS. Specific
file names and disk partitions can vary, depending on your system.
You can convert the UFS root file system on one disk to the equivalent
AdvFS root file system on a different target disk.
Requirements:
File system: root Mount directory: / Disk partition: /dev/rz1a
File system: root Mount directory: /newroot Disk partition: /dev/rz2a File
domain: root_domain Fileset: root
The AdvFS root domain is limited to one disk. Do not use the addvol command to extend the root domain.
This section presents several methods for converting the /usr
file system from UFS to AdvFS. These methods are guidelines to illustrate
the process of converting file systems to AdvFS: Specific file names, tape
drives, and disk partitions can vary, depending on your system.
Requirements:
File system: /usr Disk partition: /dev/rz3g
File system: /usr Disk partition: /dev/rz3g File domain: usr_domain Fileset: usr
Requirements:
(The file system that contains the intermediate file can be on the same
disk or on a different disk. However, do not put the intermediate file on
the /usr file system.)
File system: /usr Disk partition: /dev/rz3g Intermediate file: /tmp/usr_bck
File system: /usr Disk partition: /dev/rz3g File domain: usr_domain Fileset: usr
Use the following procedure as a guide for converting the /usr file system:
Requirements:
File system: /usr Disk partition: /dev/rz3g
File system: /usr Disk partition: /dev/rz2c Mount directory: /usr.advfs File
domain: usr_domain Fileset: usr
This section presents two different methods for converting data file
systems from UFS to AdvFS. The second method appends the first method with
additional instructions, resulting in an AdvFS file system that consists of
two independent filesets within one file domain.
The conversion methods presented here are only guidelines to illustrate
the process of converting file systems to AdvFS. Specific file names, tape
drives, and disk partitions can vary, depending on your system.
Requirements:
File system: /staff2 Mount directory: /staff2 Disk partition: /dev/rz2c
File system: /staff2 Disk partition: /dev/rz2c File domain: staff_domain Fileset: staff2
Use the following procedure as a guide for converting the /staff2 file system:
8.7 Restoring the fdmns Directory
The /etc/fdmns directory contains
a set of subdirectories, one for each file domain on your system. Each subdirectory
includes symbolic links to every volume in the file domain. AdvFS cannot
mount filesets without this directory.
Note
You must use the addvol and rmvol utilities to
add and remove volumes. Creating and removing links alone does not add or
remove a volume.
8.7.1 Restoring from Backup Media
Restoring from backup media is the preferable method for restoring the /etc/fdmns directory, provided you have a current backup copy of the
directory. You can use any standard backup facility (vdump, dump, tar, or cpio) to back up the /etc/fdmns directory. To restore the directory, use a recovery procedure that
is compatible with your backup facility.
8.7.2 Reconstructing the Directory
You can reconstruct the /etc/fdmns directory manually or
with the advscan command. The procedure for reconstructing the fdmns directory is similar for both single-volume and multivolume file
domains.
To reconstruct the two single-volume file domains, enter:
# mkdir /etc/fdmns
# mkdir /etc/fdmns/domain1
# cd /etc/fdmns/domain1
# ln -s /dev/rz1c
# mkdir /etc/fdmns/domain2
# cd /etc/fdmns/domain2
# ln -s /dev/rz2c
To reconstruct the multivolume file domain, enter:
# mkdir /etc/fdmns
# mkdir /etc/fdmns/domain1
# cd /etc/fdmns/domain1
# ln -s /dev/rz1c
# ln -s /dev/rz2c
# ln -s /dev/rz3c
Refer to the fdmns(4), mkfdmn(8),
and addvol(8) reference pages for more information.
See the advscan(8) reference page for details on using
the command to restore the /etc/fdmns directory.
8.8 Restarting the System
Unexpected shutdowns, usually as a result
of system interruption or media failure, cause you to restart your Digital UNIX
operating system. When you are forced to restart your system after an unexpected
shutdown, AdvFS is affected.
8.8.1 System Interruption
An example of a system interruption is when your site unexpectedly loses
power. This usually happens without warning.
8.8.2 Media Failure
Newer magnetic disks fail less frequently than devices based on older
technology. Nevertheless, if any single disk in a file domain fails, you
must restore all filesets in the file domain. Assuming you use the vdump command to back up your filesets, you can restore your filesets
by using the vrestore command.advfs(4), fdmns(4), vdump(8), and vrestore(8)
reference pages for more information.
8.9 Converting the root File System
Converting the root
file system to AdvFS enables booting from an AdvFS file domain and supports
AdvFS as the root file system. The AdvFS root domain must reside on a single
disk.
Assumptions:
Use the following procedure as a guide for converting the
root file system:
The converted root file system is ready to use.
# mkfdmn -r -t rz26 /dev/rz2a root_domain
# mkfset root_domain root
# mkdir /newroot
# mount -t advfs root_domain#root /newroot
# vdump 0f - / | (cd /newroot; vrestore -xf -)
# disklabel -r /dev/rrz2a > /tmp/rz2label
# disklabel -t advfs -r -R /dev/rrz2a /tmp/rz2label rz26
/dev/rz1a / ufs rw 1 1
root_domain#root / advfs rw 1 1
8.10 Converting the /usr File System from UFS to AdvFS
Relying on the fsck
utility to check and repair the /usr file system can be time-consuming.
By converting the /usr (UFS) file system to AdvFS, you can reduce
the amount of time your system is down after a system failure.
8.10.1 Using a Backup Tape to Convert the /usr File System from UFS to AdvFS
You can convert the /usr (UFS) file system to the equivalent /usr (AdvFS) file system by backing up the existing file system to tape
and restoring it to an AdvFS environment.
Assumptions:
Use the following procedure as a guide for converting the /usr file system:
# mt rewind
# cd /usr
# vdump -0 .
/dev/rz3g /usr ufs rw 1 2
usr_domain#usr /usr advfs rw
# shutdown -h now
>>> b -fl i
The system will prompt you for the name
of the kernel you want to boot. Press Return to accept the default vmunix
kernel.
# mount -u /
# mkfdmn /dev/rz3g usr_domain
# mkfset usr_domain usr
# mount -t advfs usr_domain#usr /usr
# vrestore -x -D /usr
8.10.2 Using an Intermediate File to Convert from UFS to AdvFS
You can convert the /usr (UFS) file system to the equivalent /usr (AdvFS) file system by backing up the existing file system to a
file and restoring it to an AdvFS environment.
Assumptions:
# cd /usr
# vdump -0f /tmp/usr_bck /usr
/dev/rz3g /usr ufs rw 1 2
usr_domain#usr /usr advfs rw
# shutdown -h now
>>> b -fl i
The system will prompt you for the name
of the kernel you want to boot. Press Return to accept the default vmunix
kernel.
# mount -u /
# mkfdmn /dev/rz3g usr_domain
# mkfset usr_domain usr
# mount -t advfs usr_domain#usr /usr
# vrestore -xf /tmp/usr_bck -D /usr
8.10.3 Converting from One Disk to Another Disk
You can convert the /usr (UFS) file system on one disk to
the equivalent /usr (AdvFS) file system on a different target disk.
Assumptions:
Use the following procedure as a guide for converting the /usr file system:
# mkfdmn /dev/rz2c usr_domain
# mkfset usr_domain usr
# mkdir /usr.advfs
# mount -t advfs usr_domain#usr /usr.advfs
# cd /usr
# vdump -0f - -D . | vrestore -xf - -D /usr.advfs
/dev/rz3g /usr ufs rw 1 2
usr_domain#usr /usr advfs rw
8.11 Converting a Data File System from UFS to AdvFS
By converting your
data file systems to AdvFS, you can eliminate lengthy reboots. Moreover,
you can introduce new configurations to reduce file system management overhead.
8.11.1 Using a Backup Tape to Convert a Data File System from UFS to AdvFS
You can convert a data UFS file system to the equivalent data AdvFS
file system by backing up the existing file system to tape and restoring it
to an AdvFS environment.
Assumptions:
The converted /staff2 file system is ready to use.
# mt rewind
# cd /staff2
# vdump -0 .
# umount /staff2
# mkfdmn /dev/rz2c staff_domain
# mkfset staff_domain staff2
# mount -t advfs staff_domain#staff2 /staff2
# vrestore -x -D /staff2
/dev/rz2c /staff2 ufs rw 1 2
staff_domain#staff2 /staff2 advfs rw