Digital PDFs

AA-ME93A-TE

May 1988

64 pages

Original

2.7MB

view download

OCR Version

2.2MB

view download

Document:	ULTRIX-32 Guide to System Disk Maintenance
Order Number:	AA-ME93A-TE
Revision:	0
Pages:	64
Original Filename:

OCR Text

ULTRIX-32
Guide to

System Disk Maintenance

Order No. AA-ME93A-TE

ULTRIX-32 Operating System, Version 3.0

Digital Equipment Corporation

The information in this document is subject to change without notice and should not be
construed as a commitment by Digital Equipment Corporation.
Digital Equipment Corporation
assumes no responsibility for any errors that may appear in this document.

The software described in this document is furnished under a license and may be used or
copied only in accordance with the terms of such license.

No responsibility is assumed for the use or reliability of software on equipment that is not
supplied by DIGITAL or its affiliated companies.

The following are trademarks of Digital Equipment Corporation:

DEC
DECnet

Q-bus
RT

VAX
VAXstation

DECUS

ULTRIX

VMS

MASSBUS

ULTRIX-11

MicroVAX

ULTRIX-32

ULTRIX Worksystem Software

PDP

UNIBUS

dlilaliltlal

UNIX is a registered trademark of AT&T in the USA and other countries.
IBM is a registered trademark of International Business Machines Corporation.

MICOM is a registered trademark of Micom System, Inc.
This manual was written and produced by the ULTRIX Documentation Group in Nashua, New
Hampshire.

Contents

About This Manual
AUAIENCE

cooiiiiiiiiieeieeeeeeee
e e e e e e e e e e e e e et
a et a et
—————_

OrZaniZatiOn

.......cooiiiiiiiiiiii oottt
e e et e e e e e e eeee s e e e

Related DoOCUIMEILES

vii
vii

..uuvvvviiiiiiieiieii
et et s e s e e e saasaesaenaanens

viii

CONVENLIONS .iiiiiiiiiiiieieie et
e e e e e e e e e eeeeeesseeseaeaesaeaaaraearanaes

viil

1 System Disks
1.1

Disk Organization

........ccccccceiiiiiiiieiiiiiiieeee
e esree e e e e e e eeeeeeeas

1-1

1.1.1 Disk Partition Defaults

........c.cccccvviiiviiiiiiiiiiiieeeeeeireeeeeeeeeenn.

1-2

1.1.2 Selecting Disk Partitions

........ccccoeevvvvviviiiiiiiiiieeeeeeeeiecieeeeeaeeenns

1-2

.....ccccccoviiiiiiiiiiiiiiiiiiee
et
e ee e

1-2

1.2 Paging and Swapping

1.2.1 Allocating Swap SPace

......ccccccviviiiiiiieiiiiieeiiiiiee
e eeeeveeeeeeena

1-3

1.3 File System Organization ............ccccccoooiviiiiiiiiiiieeeeeeeeeeeeeeereeseeeeeeerenns

1-4

1.3.1 Setting Up /VAT

..ovvuiiiriiiiiiieiieieieie

ee e

1-4

1.4 System DIr€Ctories ...cocveeeiiiiiiiiiiiiiiieeeeeeeee
e e e e e e eeee e e eeeeeerrreeseeeseeeans
eeee

1-6

1.5 USer DIreCtOriBs

e e e e seaaeaeas

1-7

.........ccccceeeevvevennn..
1.5.2 Sample User Directory Tree Structure ...........cccceeveveeeeevuerenn.

1-8

1.5.2.1 Clerical USEIS

1-9

...ccccccieiiiiiiiiiieieeiie ettt

e ee

1.56.1 Creating a User Directory Tree Structure

.iiuuciiiieeieee e,

1-7

2 Managing Disk Space
2.1 Monitoring File System UsSe

...ccccooovvvviiieeiiiiiiiiiiiiiiiiiiieee,

2.1.1 Checking Available Free Space
2.1.2 Checking Disk USE

2-1

........cccooeviiiiiiiiiiiiiiiiiinnnnn,

2-1

...cccoovvviiiiiiiiiiiiiiiiiiee

2-2

2.1.3 Verifying Disk QUOLAS

......oovveeiiiiiiiiiiiiiieeececeeeeeee
e

2.2 Obtaining Additional Disk Space

.....ccccceeiiiiiiiiiiiiiiiiiiiieeee,

2.2.1 Deleting Unused or Obsolete Files

2-2

2-3

...........cccccvvvvvivvrvviriviirennnnns

2-3

2.2.2 Moving a File System ......ccccooovviiiiiiiiiiiiieieiiiiiiiiceeee
e

2-3

2.2.3 Moving Files to a Larger File System

.......ccccoooevviiiiiiiiiinnnnnnnn.

.........cocccooiemmmmooieeeeeeeeeeeeeeeeeenn,

3-1

3.2 The ULTRIX Disk Partitioning Scheme .......ccccoeeevveeiiiiiiinn..

3-2

3 Disk Partitioning
3.1 Preparing to Change Disk Partitions

3.2.1 Partition Table Values

.........cccoooiiviiiviiiiiiiiiiiiiiiiiiiiiiiivineeeeen,

3.2.2 Reading the Partition Tables

........ccccocoveeviiiiiiiiiiiiiiiriiiiiiiiiin,

3.2.3 Rules for Changing Partition Tables

............ccccccovvviririrvrvvnennn.

3.3 Reviewing the Current File System Assignments

3-2
3-3

3-3

..........ccceoveveerrnen..

3-5

.........ccccoovvvvivrivrininnnnn.

3-5

3.5 Changing Partition Sizes .........ccooiiiviiiiiiiiiiiiiiiiieeeeeeeeeeeeeeeee

3-6

3.4 Reviewing the Current Disk Partition Setup

3.5.1 Changing Partition Sizes on an RA81 Disk

........cccccccrrrinnnnn.

3.5.2 Changing Partition Sizes on an RMO05 Disk (Example)

4 Bad

3-7

......

3-9

..........cccoooviiviiiiiiiiiiiiiiiieiiiiniiinenn.

4-1

Block Replacement

4.1 Automatic Bad Block Replacement

4.1.1

iv Contents

Controller-Initiated and Host-Initiated Strategies

...................

4-1

4.1.2 The Media Replacement Strategy ..............ccccccvvvrimiiiiiiennnnn.

4-2

4.2 Disk Error Messages Interpretation ..............ccccoovviiiiiiiiiiiniiiiiiniiininnnnn.

4-2

4.2.1

Logical Block NUMDEIrS

.......cccooiviiiiiiiiiiiiiiiiiiiiiiieceeieee

4.2.2 Detecting and Reporting Disk Errors

4-3

......cccccccovvvvviiiiinniiinnnnn.n.

4-3

4.2.3 Examining the Errorlog File

.......ccocooviiiiiiiiiiiiiiiniiiiiiiiiiinniiiis

4-3

4.2.4 Interpreting Error MeSSages

......cccoooooveviiiiiiiiiiiiiniiiiiiiiiieeeeevinanns

4-4

4.2.4.1 Bad Block Reported Message ............cccccevvrvvrrimriiiiniiiiiienennnnn.

4-4

4.2.4.2 More Bad Blocks Not Reported Message

.......cccooovvvvvveeennnn.

4-5

......coccccovvvvviiiiiiiiiiiiiiieeeiiiieneeiiieneees

4-5

.........cccocovvevviviiiiiiiiiiiiiiiiiiiiiiiee
e,

4-5

4.2.4.3 Transient Error MesSSAges
4.2.4.4 Forced Error Condition

4.3 Forced Error ReECOVEIY ..ooviiiiiiiiiiiiiiiie
4.3.1

Discovering Forced Errors

4.3.2 Restoring Data
4.4 The radisk UtILLY
4.4.1

........cccooovviviiiiiiiniiiiiiiieiiieeeeiieneeeens

........coooviiiiiiiiiiiiiiiee

4-6
4-6

4-6

..cooooiiiieiiii e

4.7

The Clear Option

..ccccccooveviiieiiiiiiiii
e

4.4.2 The Replace Option

4-8

...

4-8

....cccovvvviiiiiiiiiii
e

4-9

1-1:

File System Organization .........ccccccccccccceeiieiiiiiiiiiiiiiiineeiieeeeeieerreeeeeeens

1-5

1-2:

Subdirectories of /usr, One-Branch Tree

.....ccooovoveeviiiiiieeeeeeeiennn.

1-8

1-3:

Subdirectories of /usr, Three-Branch Tree

...ccoooeeeeiieeiiieeeeeeeiriieeenn..

1-9

4.4.3 The Scan OPtion

A Device

Mnemonics

Index

Figures

Contents v

Tables
4-1:

Device Support for Bad Block Strategies

A-1:

Devices Supported by MAKEDEV

vi Contents

.........coccocevvvvvvveeveeennn.

---------------------------------------------------

About This Manual

The objective of this manual is to provide you with information on
handling disk space. The guide presents background information on system

disks, describes how to partition disks, and explains dynamic bad block
replacement.

Audience
The Guide to System Disk Maintenance is written for the person
responsible for managing and maintaining an ULTRIX-32 system.

assumes that this individual is familiar with ULTRIX-32 commands, the
system configuration, the system’s controller/drive unit number assignments

and naming conventions, and an editor such as vi(1)

or ed(1).

You do

not need to be a programmer to use this guide.

Organization
This manual consists of four chapters, an appendix and an index.
The chapters are:

Chapter 1:

System Disks
Provides an overview on disks and describes default
values for system disks.

Chapter 2:

Managing Disk Space
Explains how to check disk space and use, and how to
obtain additional space on a disk.

Chapter 3:

Disk Partitioning
Explains how to partition disks using the chpt

command.
Chapter 4:

Bad Block Replacement
Describes how the system replaces bad blocks and
explains how you can force bad block replacement, if
necessary.

Documents

You should have the hardware documentation for your system and
peripherals.

Conventions
The following conventions are used in this manual:
special

In text, each mention of a specific command, option,
partition, pathname, directory, or file is presented in this
type.

command(x)

In text, cross-references to the command documentation
include the section number in the reference manual where

the commands are documented.
cat(1) command.

For example: See the

This indicates that you can find the

material on the cat command in Section 1 of the reference
pages.

literal

In syntax descriptions, this type indicates terms that are

constant and must be typed just as they are presented.
italics

In syntax descriptions, this type indicates terms that are

variable.

[]

In syntax descriptions, square brackets indicate terms that
are optional.
In syntax descriptions, a horizontal ellipsis indicates that
the preceding item can be repeated one or more times.

function

In function definitions, the function itself is shown in this
type. The function arguments are shown in italics.

UPPERCASE

The ULTRIX-32 system differentiates between lowercase
and uppercase characters.

Enter uppercase characters only

where specifically indicated by an example or a syntax line.
example

In examples, computer output text is printed in this type.

example

In examples, user input is printed in this bold type.

This is the default user prompt in multiuser mode.

This is the default superuser prompt.

>>2>

This is the console subsystem prompt.

viii About This Manual

In examples, a vertical ellipsis indicates that not all of the
lines of the example are shown.

In examples, a word or abbreviation in angle brackets
indicates that you must press the named key on the
terminal keyboard.

In examples, symbols like this indicate that you must hold

down the CTRL key while you type the key that follows
the slash.

Use of this combination of keys may appear on

your terminal screen as the letter preceded by the

circumflex character.

In some instances, it may not appear

at all.

About This Manual ix

System Disks 1

This chapter provides background information about the defaults established
on your system disk when you perform a basic installation. If you change
your system disk defaults during an advanced installation or by using the

chpt command, some of the information in this chapter may no longer

apply.

Refer to chpt(8) in the ULTRIX Reference Pages and Chapter 3 of
this manual for more information on changing disk partitions.
The following sections describe:

Disk organization

Disk partition defaults

Selecting disk partitions

Paging and swapping

Allocating swap space

File system organization

Setting up /var

System directories

User directories

1.1

Disk Organization

A disk consists of storage units called sectors. A disk sector is usually 512
bytes, and sectors are grouped into a maximum of eight partitions.
However, not all disks contain the same number of partitions or the same
partition sizes.
See ra(4,) hp(4,) and rd(4) in the ULTRIX Reference

Pages for a listing of supported disks and partition sizes.
Disk partitions form logical disk boundaries that separate each file system.
Each partition can contain one file system, and several file systems can
reside on the same disk.
A file system can be smaller but not larger than

the partition size.

1.1.1

Disk Partition Defaults

When you perform a basic installation, the install procedure establishes
certain disk partition defaults for you.
®

For example:

Partition a of the system disk (usually drive 0)

contains the root file

system.

Partition b of the system disk is reserved for use as a paging and
swapping area, and for crash dumps.

Partition g of most system disks is reserved for the /usr file system:.

(On RA82 and RA90 disks, the default area for /usr is partition d.)
If partition g of the system disk is too small to contain the /usr file
system, some subdirectories of /usr can be placed on another partition
or, if available, on another disk.
o

Partition h of the system disk is reserved for the /usr/users file

system.

However, if the system disk does not have a partition h,

then the default is to put /usr/users on partition g (subordinate to
the /usr file system).

1.1.2

Selecting Disk Partitions

By selecting the file systems to be placed in each partition, you can
monitor growth and activity of the disk.

If you have multiple disks, you

can divide users across the disks for better input and output response. You
can also accommodate growth by placing the file systems you expect to

expand on partitions that are large enough to accommodate that growth.

See Chapter 3 of this manual for details on selecting and changing disk
partitions.

1.2

Paging and Swapping

The kernel needs disk space for paging and swapping to provide a timesharing environment and virtual memory to VAX users. Swapping occurs
when there is not enough physical memory for a process or processes in
the system. Paging occurs to bring data into memory, to copy data out of
memory, or to accommodate other processes that require physical memory.

Programs do not have to be written to fit within certain boundaries of
physical memory.

called pages.

The hardware is designed to process programs in lengths

If a particular page of memory is not modified during

processing, it is overwritten when that memory is needed for something

else.

When the page of memory is needed again, it can be read in from

the original file on disk.

If a page of memory is modified during

processing, it is written out to disk so that the copy on disk reflects the

page in memory.

1-2 System Disks

This writing and reading of modified pages of memory is

one of the ways the system uses the paging and swapping area on disk.
The kernel keeps as many processes (programs being executed by users)

physical memory as it can, and lets each one execute during its own
particular time slice.

Swapping occurs when a new process needs to be

loaded and executed, but there is no room for it in memory.

An old

process is swapped out to disk, to wait its turn for processing again.

When that time comes, the process is swapped into memory to continue.

1.2.1

Allocating Swap Space

The area reserved for swapping should be substantially larger than the total
available physical memory in the system.

A rule of thumb is to give at

least two times the amount of swap space as you have physical memory. If
the swap area is too small, the system will not be able to make use of all
the available physical memory.

If you receive messages indicating there is

not enough memory, you probably need a larger partition size for the swap
area.

By default, the installation procedure automatically allocates partition b of
the system disk for paging and swapping.

If you determine that you need

more paging and swapping space, you could use a different partition.

this is impossible, you could change the size of partition b on the system
disk by using the chpt command as described in Chapter 3.

usually need to change the partitions of the disk.

You do not

The existing disk

partitions will generally meet the needs of your system’s users.

You can also use a second partition on a second disk for paging and
swapping.

In this case, the kernel interleaves its paging and swapping

operations between the two disks, and your system can run faster than it
would with only one disk.

This is done by making the necessary

assignments in the configuration file.

For information on how to define

this configuration, refer to the Guide to System Configuration File
Maintenance.
Local paging and swapping is available to clients who have a local disk and
are operating in a diskless environment.

Refer to the Guide to Diskless

Management Services for details.
Note

Avoid selecting partition a of any disk for use as a swap
partition.

Partition information for the entire disk resides in the

superblock of the a partition’s file system.

If a customized

partition table has previously been defined for the disk, the

information will be destroyed when other data is swapped to
partition a.

System Disks 1-3

1.3

File System Organization

ULTRIX-32 supports two file systems:
the ULTRIX File System (UFS),
and the Network File System (NFS).
This section describes the ULTRIX
file system.
For a discussion of NFS, see the Guide to the Network File
System.
The root file system is divided into sharable data, /usr, and nonsharable
data, root (/).

This separation by data use enables server systems to share

the /usr file system efficiently with client systems.
The file systems are further divided into the following content areas: static

data files, variable data files, and executable data files.
|

Static data files remain fairly constant over time.

Sharable static

data, like library routines, documentation macros, and uucp files are
in /usr/lib which is symbolically linked to /lib.

The /usr/etc directory

contains sharable system maintenance tools and daemons.
Nonsharable, machine specific data files, such as aliases, crontab, and
sendmail.cf are in /etc.

Variable data files are dynamic files that change in size,

such as

spool files.

On server

Variable data files reside in /var or /usr/var.

systems where the user files are shared with diskless clients, all
variable data files should be in /var, and /var should be set up as a
separate file system.
e

Executable data files are machine architecture-specific files.
select set of executables can be shared by most systems.

All but a
Shared

executables, such as ls, mail, and pwd are in /usr/bin. Single user
executables, such as init and fsck are in /bin.
The file system organization simplifies system management by placing
volatile files of the same type in the same directories.
content simplifies the system management task.

Separating files by

Unlike the area reserved

for variable data, the areas reserved for executables or static data will not
change in size. This organization also greatly improves system debugging,
since all the logging files are contained in the variable areas.

1.3.1

Setting Up /var

When you perform the basic installation, there are only two file systems, /

and /usr.

Because of the space limitations in this configuration, /var does

not exist.

Files and directories that would normally reside in /var are

located in /usr/var, which is symbolically linked to /var.
When you perform an advanced installation, the software asks you if you
want to set up a separate file system for /var.

If you also set up

/usr/users as a separate file system, you can mount /usr as a read only file
system.

This file system organization is preferred if you are setting up a

1-4 System Disks

system that will share /usr.

It provides the most secure access to shared

files and simplifies system management tasks by limiting extremely

dynamic files to one area.

Non-sharable
root (/)

xecutable
/bin

achilippe

speciiic
/ete

Static
/lib

Variable
Jvar

xecutable

/bin

Sharable
lusr

aemons

and Tools
lete

ZK-0014U-GE

Figure

1-1: File

System Organization

System Disks 1-5

1.4

System Directories

In addition to the directories shown in Figure 1-1, the system needs
certain directories for standard operations.

The following list describes

these directories, their use, and - where relevant - the recommended space
allocation.
°

The /var/adm directory contains data files generated by administrative
programs such as system accounting and the error logger.

The data

and files in /var/adm can vary widely across systems and over time.

The /var/adm/acct file, for example, can easily grow by 5 Kbytes a
day.

In addition, /var/adm is the default location for the system

crash directory, /var/adm/crash.
We recommend that /var/adm have a partition size equivalent to
roughly two times the amount of memory on your system.
°

The /var/spool directory contains files being held for spooling.

For

example, files for the line printer, mail messages, or network file
transfers are held in this directory.
We recommend that you monitor the variable growth of /var/spool to

determine the amount of space needed at your site.

If your users

place heavy demands on these facilities, then allocate space
accordingly.
®

The /tmp directory is used by various system and user programs for
temporary files.

For example, the vi editor creates a temporary file

in the /tmp directory.

The /usr/etc directory contains sharable system maintenance tools and

daemons, such as config, talkd, and nfsd.
o

The /usr/hosts directory contains a file for each node on the local
area network.

The /usr/hosts/MAKEHOSTS command creates these

files, which are linked to /bin/rsh, the remote shell program.

Each of

these files typically requires one Kbyte of disk space.
®

The /usr/man directory contains the source for documents printed in

the ULTRIX Reference Pages.
°

The /usr/users is the default for the home directory for each of the
system users.

In an environment where /usr is shared with client

systems, /usr/users should be set up as a separate file system.

you installed your system with the Advanced Installation, you were

provided with the option of creating a separate file system for
/usriusers.

See the following section, User Directories, for more

information on this file system.

1-6 System Disks

1.5

User Directories

By default, the login (home)
/usrfusers.

directory for users is subordinate to

You can change this default so /usr/users contains one or more

subdirectories for your users. This section explains how to reorganize the
/usr/users file system according to the requirements of your users.
In general, any user who logs in to the system has access to the files in
his home directory.

On local systems, the default locations for individual home directories
are in subdirectories of /usr/users.

For example, a user named

mjadams could have a home directory with the pathname,

/usr/users/mjadams.
°

On a remote client system that shares the /usr/users file system,
users can mount the home directory on any established mount point.

On client systems that run Yellow Pages, the home directory mount
point on the client should have the same pathname as the pathname
for the home directory on the server system. For example, if the
pathname of the home directory, rjones, is /usr/users/rjones, you must
create a directory for /usr/users/rjones on your client system as the
mount point.

Because Yellow Pages systems share the password file,

the home directory pathname, userid, and groupid must match on the
client and the server systems.

Refer to the Guide to the Yellow

Pages Service for more information on Yellow Pages.

1.5.1

Creating a

User Directory Tree Structure

Changing the default structure requires that you know how to set up
directory tree structures and that you understand the needs of your users.
You can create a directory tree structure for users in the /usr directory
with as few or as many branches as you want.

Some sites, for example,

create home directories for all users within one subdirectory, as shown in

Figure 1-2.

If the space required by all your users fits on one disk

partition, there is no reason to divide the directories. Other sites may
require more complex allocations, as shown in Figure 1-3.

If you need

more space for /usr/users, you can move the file system to another

partition.

(For more information on moving file systems, see Chapter 2.)

System Disks 1-7

/usr

{

/users
.

/adams

/ Jones

/smlthJ anblonsky
ZK-0015U-GE

Figure

1-2: Subdirectories of /usr, One-Branch Tree

1.5.2

Sample User Directory Tree Structure

This section explains how you can organize user directories into three
subdirectories for three types of users, all with different requirements.

This example is illustrated in Figure 1-3.
In this example, users are divided into three categories according to disk
use:

clerical, scientific, and administrative.

1-8 System Disks

/usr

/clerical

/sclentific

(/Jln J (/nomyj (/Fred J

/o.cmin

(/allcej fi/donJ (/ Je{-‘FJ

(/harryJ (/somuh) (/mkeJ
ZK-0016U-GE
Figure

1.5.2.1

1-3: Subdirectories of /usr, Three-Branch Tree

Clerical Users -

Assume that there are 10 clerical workers on

the system, and they each have at least 25 files on hand at any given
time.
These users generate nroff source and output files for memos,
papers, and reports.

30 Kbytes each.

We will assume that these files require an average of

The average load, in this case, is 7.5 Mbytes, which you could put on an
8-Mbyte partition.

To allow for growth and periods of high production, you
could enlarge the size of an 8-Mbyte partition or allocate the directory to a
larger partition for this group.

System Disks 1-9

1.5.2.2
Scientific Users - Assume there are 15 graduate students in the
group of scientific users and they generate C program files — source,
object, and executable.
These users keep an average of 10 modest-sized
programs (20 Kbytes for source, 10 Kbytes for object, and 15 Kbytes for
executable programs), totaling 6.75 Mbytes.
Also assume there is one
professor who keeps 10 of his programs and data bases around at any
given time (he generates huge data files of about 1,000 Kbytes); he would
require around 11.2 Mbytes.
The total requirement for the scientific staff
would be around 18 Mbytes.
Allowing for growth, the scientific staff might
require a medium-sized partition on a large disk (partition d or € on an
RAS81 disk, for example).
1.5.2.3

Administrative Users -

The administrative staff’s needs are also

fairly large, but they have a special security requirement.
salary and personnel files that must remain private.

They keep

To maintain this

privacy, they should be assigned a partition on a disk with removable
packs.

Use the newfs command to create new file systems on partitions that are
large enough to hold them and any anticipated growth.
in the ULTRIX Reference Pages.

Refer to newfs(8)

You should keep in mind, however, that

the size of a file system is defined by the size of the partition on which it
resides.

If you put a small file system into an excessively large partition,

the file system would be wasting the excess disk space on that partition.
For example, on RP07 disks, you might have mounted a large file system
on partition g because this partition overlaps partitions d, e, and f.

using partition g, the file system will have more room than if you use
partition d, e, or f.
chpt program.

To review the current disk partition setup, use the

See Chapter 3 of this manual and the ULTRIX Reference

Pages for details on the chpt command.

1-10 System Disks

Managing Disk Space 2

This chapter discusses methods of monitoring file system use and managing
It explains how to check the disk space and use and

file system data.

describes different methods you can use to obtain additional disk space.

2.1

Monitoring File System

Use

To ensure an adequate amount of free disk space, regularly monitor the
disk use of your configured file system.
To ensure adequate free disk
space:

Check available free space using df

Check disk use using du -s

Verify disk quotas (if imposed)

2.1.1

Checking Available

:
using quot

Free Space

To ensure sufficient free space for your configured file systems, you should
regularly check the amount of free disk space available in all of the
mounted file systems.
To see how much free disk space there is, use the

df(1) command.
This command reports the amount of free space available
on all of the currently mounted file systems.
When invoked, the df
command generates a listing similar to the following:
#

Filesystem

total

kbytes

node

kbytes

used

free

used

7429

6141

546

92%

/dev/rala

percent
Mounted

/dev/rala

7429

437

6250

/tmp

/dev/ralg

175015

143520

13994

91%

/usr

/dev/rale

30519

14342

13126

52%

/dev/ralh

/usr/spool

313233

134316

147594

48%

/usr/staff

For each file system, the df command reports the file system’s configured
size (Kbytes), the amount presently used, the amount presently available
(free), the percentage used, and the directory on which the file system is

mounted.

You can also use the df command with the —i option to display both the
free and the used inodes.
its options, see df(1)

For more information on the df command and

in the ULTRIX Reference Pages.

File systems are usually configured with a minimum percentage of free
space established.

When constructing a new file system, the newfs

command reserves a minimum percentage of the space (the default is 10

percent).

This default percentage allows for a report in excess of 100

percent.

When interpreting the free space report, therefore, you should

look for significant changes in file system disk use.

A file system is

considered to have insufficient space when the percentage used exceeds 90.

2.1.2

Checking Disk Use

After determining that a file system has insufficient space available, check
how its space is being used. The du program pinpoints disk space

allocation by user.

With this information you can decide who is using the

most space and who should free up disk space.
To display a summary of how space is being used on a file system, use
the du command with the —s option specified.

The following example

shows how to display a summary of the space use by all subdirectories

(accounts)
#

in the /usr/users file system:
-s

/usr/users/*

This command displays a summary of the number of blocks used by each
main subdirectory in the specified file system (/usr/users).

Normally, this

information is sufficient to determine which users have the most disk
space.

You can also use the quot command, to list the number of blocks in the
named file system owned by each user.

disk space use, refer to quot(8)

For more information on checking

and du(l)

in the ULTRIX Reference

Pages.

2.1.3

Verifying Disk Quotas

If you are enforcing user disk quotas, you should verify your quota system
periodically.

You can use the following commands to compare the

established limits with actual use:

quot, quotacheck, and repquota.

The quot, command displays the actual block use for each user.

The

quotacheck command verifies that the actual block use is consistent with

established limits.

The repquota command displays both the actual disk

use and the established limits.

2-2 Managing

Disk Space

If you find it necessary to change the established quotas, use the edquota
command.

This command allows you to set or change the limits for each

user.

Note

To enable automatic quotas, set the third field of the /etc/fstab
file to rq for the listed file system.
For further information on disk quotas refer to the

ULTRIX-32

Supplementary Documents, as well edquota(8), quotacheck(8), quotaon(8),
quotaoff(8), and repquota(8)

2.2

in the ULTRIX Reference Pages.

Obtaining Additional Disk Space

Once you have checked disk space and use, you may need to establish

additional space on the disk or on a file system.

You can use the

following methods, depending on your system and the needs of its users:
o

Delete unused or obsolete files

Move a file system

Move files to another file system

Change partition tables

A complete discussion of changing partition tables is contained in Chapter
3.

2.21

Deleting Unused or Obsolete

Files

You should request that your system users remove any unused or obsolete
files.

If there is still an insufficient amount of free space, request that
users of that file system dump their infrequently used files.

For further information on backing up and restoring files, see Guide to
System Backup and Restore.

2.2.2

Moving a

File

System

When moving a file system from one disk pack to another, follow these
steps:

Dump the file system (level 0)

with the dump command.

Create a clean file system with the newfs command.

Restore the file system with the restore command.

Managing

Disk Space 2-3

The Guide to System Backup and Restore contains information on backup
and restore procedures for your system.

2.2.3

Moving Files to a

Larger File System

There are two methods of moving files to a larger file system:
the files and merging individual files.

moving all

However, before you move any files,

make a level 0 dump of the entire file system.
When transfering all the files to a larger file system, restore the level 0
dump on the larger target file system.
When merging individual files from one file system into another, create the
appropriate file and restore it to the target file system.
restore individual files, use the tar command.

To create and

For example, to create the

file named file1, type:
#

tar

filel

In this example, the tar command creates filel

on the default output

media, usually the default tape device (dev/rmtOh).

Then, after creating files on the default output media, use the tar
command with the xp options to restore (extract)
device to the target file system.
#

tar

the file from the tape

For example,

filel

This command extracts filel

from the default tape device to the current

working directory using the pathname on the tar archive media.

The p is

a superuser option that causes tar to use the original protection code
assigned to file1.

2-4

Managing

Disk Space

Disk Partitioning 3

This chapter provides the information you need to change the partition
sizes of your system disks.

In general, you allocate disk space during the

initial installation or when adding disks to your configuration.

However,

there are cases when it is necessary to change the partition sizes on your
system disks in order to accommodate changes at your site and to improve
system performance.

The following sections describe:
o

Basic planning and preparation

The disk partitioning scheme used by ULTRIX systems

The chpt(8) program

Step by step procedures and examples

For a review of the concepts pertaining to ULTRIX system disks, see
Chapter 1 of this manual.

3.1

Preparing to Change

Disk Partitions

Before making any changes to the partitions on your system disks, you

should:
1.

Assess the number of file systems you need and the amount of space
each requires, including potential growth space.

Here, you might ask:

Are users logically and optimally grouped within /usr/users?

Are other files and directories logically and optimally grouped
within file systems?

Does this setup satisfy our current needs?

Does this setup satisfy our projected needs?

Analyze the statistics available to you regarding current system
performance.
o

Here you might ask:

Does the current disk partitioning setup optimize the average

and peak demands of system users?
°
3.

Would a different setup improve system performance?

Review the current disk partition setup and file system allocation.

For example, you should know:

The device type and partition defaults for all disks at your site

The current size of each partition on the disk(s)

The current location of each file system on the disk(s)

Back up all files systems that exist on a disk that you intend to

reconfigure.

3.2

The

ULTRIX Disk Partitioning Scheme

Each device driver on the system contains a set of default partition tables
with one table for each type of disk on the system.

This set of tables is

the same for all ULTRIX systems:
o

For all disks, the default partition table is copied into the active
partition table for that particular disk when the disk is first opened,
or whenever the disk volume is changed (for example, the drive is
turned off and on).

For some disk packs, the default partition tables are copied into their
respective active partition tables every time the disk is opened.

The active set of partition tables contains one table for each disk

configured into the system.

3.2.1

Partition Table Values

The ULTRIX partitioning scheme allows you to use the values for the
default partition tables or to create or modify one or all of the partition
table values for individual disks.
You can continue to use the default
values in the active partition tables in the device driver, or you can modify

the partition table for a particular disk.

There are two ways to override

the default values:
1.
2.

Use the chpt(8) command
Mount a disk pack which already contains a different partition table
in the superblock of its a partition.

3-2 Disk Partitioning

For example, making a new partition table allows you to have two RMO05
disks on the system with different partitions, one using the default values
for all RMO5 disks, the other using its own partition table, which is set up
according to your space needs.

You can port the modified disk to any

other ULTRIX system, or you can modify the partition table for particular
disk.

3.2.2

Reading the

Partition Tables

The first time a disk is opened, the device driver copies the disk’s default

table in the driver that is to be used.

The device driver then checks the

superblock of the disk’s a partition to see if the disk has an existing
partition table.

If there is no partition table, the driver uses the default

values.
If the device driver finds a partition table in the superblock, the driver

copies that table into the disk’s active partition table in the device driver.

This overwrites the default values that were originally copied into the
active table.

The active set of partition tables in the device driver are the ones that
the system uses for each disk, whether a partition table exists in the
disk’s superblock or not.

Remember, the active set of partition tables in

the device driver contains one table for each disk on the system for that
driver.

Each table is a copy of either the default table in the driver or

the individual partition table in the superblock of the a partition of the
disk.

There are two reasons why a disk might not have its own partition table:
1.

The a or ¢ partition does not have a file system for a partition table

to reside in.
2.

3.2.3

An individual partition table was never created for the disk.

Rules for Changing Partition Tables

Rules and guidelines to follow when changing disk partitions are:
°

You must have superuser privileges to use chpt.

Back up all the file systems before changing the tables if there is

any data on the disk.

The new partition will overwrite the old file

systems, destroying the data.

Disk Partitioning 3-3

You cannot change the offset (beginning sector) or shrink any
partition on a mounted file system or on a file system that has an
open file descriptor.
If you need only one partition on the entire disk, use the existing
partition c.

A disk’s partition table always resides in the superblock of partition
a.
Therefore, you must have an a partition with a file system on it

before you can change the tables.
If you plan to change the partitions other than a, and there is no
file system in partition a, create a file system on partition a before
changing any others (use newfs to create a file system).
Partition a
must begin at the start of the disk (sector 0).

If your disk is divided into cylinder numbers, translate them into
sector numbers before using the change feature of the chpt command

( —p).

Here is the basic calculation:

(No.

sect/track)

(No.

sect/cyl)

(cyl

(No.

tracks/cyl)

No.)

(No.

sect/cyl)

sect

For example, suppose you have an RMO05 disk and you need the
beginning sector number of the fifth cylinder.
You can find the
number of sectors per track and tracks per cylinder by using the
diskpart(8) command:
#

/etc/diskpart

rm05:

rmO5

#sectors/tracks=51,

#tracks/cylinders=14

#icylinders=1248
partition

size
15884

33440

81
822

500384

15884

562

588

55936

589

680

86176

681

825

158528

562

822

291346

561

Now you can perform the calculation:
51
714

range

714

3570

3-4 Disk Partitioning

Note
Be careful when changing partition tables; you could
overwrite data on the file systems or make the system
inefficient.

If the partition tables become corrupted while

you are changing the partition sizes, you can return to the
default partition table with the —d option of the chpt

command.

3.3

Reviewing the Current File System Assignments

Before you change the partition tables or reallocate disk space, review the
current file system assignments on your disk.

To do this, use the df

command.
The df command displays the current file system assignments, showing the
partitions and space being used by the system.

In addition, it displays the

free space available on all file systems listed in /etc/fstab, and lists the
partitions for each disk.
For example, if you have two file systems mounted on an RA81

system

disk, when you use the df command, you will see something like this:
#

Filesystem
node

total

kbytes

percent

kbytes

used

free

used

Mounted

/dev/rala

7429

5371

1315

80%

/dev/ralh

388950

118000

270950

30%

/ustr

As this example shows, the df command gives you a broad view of the

disk partition assignments that have been made.

Note that any disk

without file systems mounted on it does not show in the df display.
Refer to the ULTRIX Reference Pages for more information on df(1).

3.4

Reviewing the

Current Disk Partition Setup

Before changing the size of a disk partition, review the current assignment.

To do this, log in as superuser and use this format:
# chpt -q device
This command invokes the chpt program.

Including the —q option

indicates that you want to view the partition sizes.

The device variable

indicates the specific disk you are checking.

Specify the device with its directory name (/dev), followed by the raw
device name and partition a or ¢.

For example, to check current partition

sizes on an RA81, type:
#

chpt

/dev/rra0a

Disk Partitioning

3-5

The chpt program responds with output similar to this:

Current partition table:
bottom

th
O 0.0

partition

3.5

top

size

overlap

15883

15884

49323

33440

891071

891072

c
c,h

a,b,d,ef,g,h

340670

356553

15884

h,c

356554

412489

55936

h,c

412490

891071

478582

h,c

49324

131403

82080

131404

891071

759668

c,d,e,f

Changing Partition Sizes

To change disk partition sizes with the chpt command, follow these steps:
1.

Display the current partition table.

Unmount any active file systems.

Back up all data on the disk.

Calculate the new partition parameters.

Change the partition parameters.

Make and mount file systems as necessary.

Restore any backed up files to the new partitions.

Check the consistency of the new file system.

When following these steps, use the chpt command, which lets you change
the partition table of an individual disk without rebuilding the kernel and

rebooting the system.
The chpt format is:
letc/chpt ~a —-q -d -v —px offset size ... device

For more information on changing partition sizes, see chpt(8)

in the

ULTRIX Reference Pages.
The following sections explain the procedure and give examples for:
o

Changing partition sizes on an RA81 disk

Changing partition sizes on an RMO05 disk

3-6

Disk Partitioning

Changing Partition Sizes on an RA81

3.5.1

Disk

The following example shows the correct way to change a partition table on
a running system.

It then shows what happens if you try to change

partitions incorrectly so you can avoid these mistakes.

Assume that you have an RA81 disk on drive 7 of your system and that
you want to customize it to your system’s needs by expanding partition a

to include all the space in partition b.

Step 1: Display the Current Partition Table
Use the chpt command with the

g option to see the status of the

partition table:
#

chpt

/dev/rra7a

Current

partition

table:

bottom

top

size

15883

15884

49323

33440

c,h

overlap

891071

891072

a,b,d,e,f,g,h

340670

356553

15884

c,h

356554

412489

55936

c,h

412490

891071

478582

<¢,h

49324

131403

82080

131404

891071

759668

«c¢,d,e,f

Step 2: Unmount Active File Systems
Use the mount command to see which file systems are mounted on which

devices:
#

/etc/mount

The mount program responds with information such as:
hpOg

hpOh

/usr/staff

raZa

/mnt

/usr

From this response you learn that the /mnt file system is mounted on
partition a of the RA81 disk. This is the disk where you want to change
partition sizes.
#

Unmount /mnt as follows:

/etc/umount

/dev/ra7a

Disk Partitioning

3-7

Step 3: Back Up the Disk
You should back up the entire disk before changing partition sizes.
In this example no files need to be backed up.

If, however, files did need

to be backed up, you would use the dump command to back up the files
to another media.

Step 4: Calculate the New Partition Parameters
This step shows how to expand partition a of an RA81 disk so that it
encompasses partition b.

By looking at the partition table generated by

the chpt command, you can see that you need to extend the size of
partition a to the ending sector of partition b.
you

can

Here is the formula that

use:

(new

length

(beginning

(size

Here is the calculation:
49324

15884

33440

Step 5: Change the Partition Parameters

Shown here are three examples of changing partitions in an RA81 pack.
Method A shows how to use the chpt command correctly with the -b
option.

Methods B and C show how not to change partitions.

Method A -

Correct -

an active partition.

Here is a correct way to extend the length of

This example shows how to extend the a

partition by the size of partition b:
#

chpt

-v

-pa

49324

/dev/rraZ7a

/dev/rra7a
Current

partition

table:

bottom

top

size

49323

49324

overlap
b,c

15884

49323

33440

c,h

891071

891072

a,b,d,e,f,g,h

340670

356553

15884

c,h

356554

412489

55936

c,h

412490

891071

478582

c¢,h

49324

131403

82080

131404

891071

759668

c¢,d,e,f

3-8 Disk Partitioning

Method B -

Incorrect -

One of the rules for changing partitions is

that the chpt command does not let you shrink an active (mounted)
partition.
#

Here is what happens if you try:
0

800

/dev/rra7a

chpt:

chpt

cannot

-pa

set

partition

Mount

device

busy

Method C -

Incorrect —

chpt

-pa

driver:

Another rule is that you can never change

the starting sector (offset)
#

table

of partition a:

800

/dev/rra7a

chpt:

cannot

set

partition

Mount

device

busy

table

driver:

Step 6: Make and Mount File Systems
Then, place the file system on the partition, using the newfs command.
#

/etc/newfs

/dev/rra7a

Warning:

partition

Warning:

656

/dev/rra’7a:
51

sectors

(16

c¢/g,

25.3Mb

5.85Mb/g,

superblock

32,

sector(s)

49324

backups

11520,

23008,

ra8l

table

overriding

last

sectors
in

cyl

unallocated

cylinders

tracks,

groups

2048

1/g)

(for

fsck

34496,

/etc/disktab

cylinder

-b#)

at:

45984,

If you are ready to use the file system, remount it on the partition with
the mount command:
#

/etc/mount

/dev/ra7a

/mnt

Step 7: Restore Backed Up Files

In our example, there are no files to restore.

If there were, you would

use the restore command.

Step 8:

Check File Consistency

Check the consistency of the restored files using the fsck command.
3.5.2

Changing Partition Sizes on an RMO05 Disk ( Example)

Although this example is for an RMO05 disk, the principles of this example
apply to any disk that begins each partition at the beginning of a sector.
Here is a scenario that requires changes to the partition table.

Assume

that you have an RMO5 disk on your system and you want to customize it

Disk Partitioning 3-9

to your system’s needs.

Assume your system requires twice as much swap

space as the default partition table allows.

the size of partition b.

Therefore, you need to double

One way to extend the length of partition b is by

using space in the beginning of partition h.

For this example, assume that the RMO05 disk has just been configured
into the system and that no data is yet on the disk.

Therefore, there is

no need to back up or restore any files.
Keep in mind that RMO5 disks always begin each partition at the

beginning of a cylinder.

Therefore, the starting sector of one partition may

not immediately follow the last sector of the previous partition.

This

creates unaddressable disk space.

On an RMO5 disk, there are 32 sectors per track and 19 tracks.
there are 608 sectors per cylinder (32*19).
Here is the default table, found in hp(4)
Disk

Start

Length

for an RMO05 on drive O:

Cylinders

hpOa

15884

0-26

hpOb

16416

33440

27-81

hpOc

500384

0-822

hp0Od

341696

15884

562-588

hpOe

358112

55936

589-680

hpOf

414048

86240

681-822

hpOg

341696

158592

562-822

hpOh

49856

291346

82-561

Here is how you want the partitions to be divided.

The lengths of the

partitions being modified are highlighted:
Disk

Start

End

Length

Cylinders

hpOa

15884

0-26

hpOb

16416

83295

66880*

27-136

hpOc

500384

0-822

hp0d

341696

357579

15884

562-588
589-680

hpOe

358112

414047

55936

hpOf

414048

500287

86240

681-822

hpOg

341696

500287

158592

562-822

hpOh

83296

341202

257906*

137-561

3-10 Disk Partitioning

Thus

There are 823 cylinders.

Now that you have determined what changes are necessary, follow these
steps:

Step 1: Display the Current Partition Table
You must know how the disk is currently partitioned. To view the current

table, use the chpt command, as follows:
#

/etc/chpt

g¢q

/dev/rhpOa

/dev/rhpOa
No

partition

using

table

default

Current

partition

found

table

from

superblock.
..

device

driver.

table:

bottom

top

size

15883

15884

overlap
C

16416

49855

33440

c,h

500383

500384

a,b,d,e,f,g.,h

341696

357579

15884

c,g

358112

414047

55936

c,g

414048

50028

86240

c,g

341696

500287

158592

.c¢,d,e,f

49856

341201

291346

In this example, chpt indicates that there is no partition table in the
superblock of partition a.

Therefore, the chpt command displays the

partition table found in the device driver for the RMO05 disk on drive O.
Before you can change the partition table, you must copy the RMO5 table
in the driver into the superblock of partition a of the disk.

The —a option

of the chpt command does this for you:
#

chpt

-a

/dev/rhpOa

Step 2: Unmount Active

File Systems

Use the mount command to check for mounted file systems.
In this example, a new RMO05 disk is being added to the system.
Therefore, you know there are no file systems mounted on any partitions.

If a file system did exist, you would unmount it using the umount
command.

Step 3: Back Up File Systems
You should back up any file systems existing on the partitions to be
changed.

(For data protection, it is recommended that you back up all

data on the disk before changing partition sizes.)
Because the example calls for adding a new RMO5 disk to the system,
there is no data on the disk.

Therefore, in this example, you do not need

to back up any files.

Disk Partitioning 3-11

Step 4: Calculate the New Partition Parameters
To calculate the new partition parameters, use the sector numbers (size)
that the chpt

q command provided.

You can do your calculations on a

piece of scratch paper, with a calculator, or on the system.

The following

example shows the formula for doubling the size of partition b and

reducing the size of partition h to reflect the new size of partition b.

The

formula for this example is:
(new

size

(beginning
(size

(current

size

(size

new

(beginning

size

(beginning

sector

(ending

sector

2
of

b)
of

The calculation for this example is:
66880

33440

83296

66880

2
16416

257906

341201

83297

You now have all the necessary data for changing partitions b and h.

Step 5: Change the Partition Table Parameters
First, use the chpt command to double the size of partition b:
#

chpt

-v

—-pb

16416

66880

/dev/rhpOa

/dev/rhpQOa
New

partition

table:
bottom

top

size

15883

15884

overlap
o)

16416

83295

66880

c,h

500383

500384

a,b,d,e,f,g,h

341696

357579

15884

c,g,h

358112

414047

55936

c,g,h

414048

500287

86240

c,g,h

341696

500287

158592

¢c¢,d,e,f

49856

341201

291346

Second, use the chpt command to decrease partition h by the size that
partition b was expanded:

3-12

Disk Partitioning

+1)

chpt

-v

-ph

83296

257906

/dev/rhpOa

New

partition

table:
bottom

top

size

15883

15884

16416

83295

66880

c,h

500383

500384

a,b,d,e,f,g,h

341696

357579

15884

c,g,h

358112

414047

55936

c,g.h

414048

500287

86240

c,g,h

overlap

341696

500287

158592

c¢,d,e,f

83296

341201

257906

You have now changed the partition table.

Step 6: Make and Mount File Systems
Before you can mount file systems on the partitions, you must make the
necessary file systems using the newfs command.

Make new file systems

for all partitions whose bottom addresses you changed with the chpt
command.

In this example, because b is swap space, you only need to

make a new file system for partition h:
#

/etc/newfs

/dev/rhpOh

rm05

Now you can mount any necessary file systems on the various partitions.

Step 7: Restore Backed Up Files
If you backed up any files before changing the partition tables, restore
those files to their proper file systems using the restore command.
In this example, no files were backed up so there are no files to restore.

Disk Partitioning 3-13

Bad Block Replacement 4

This chapter describes the replacement of bad blocks on DIGITAL Storage
Architecture (DSA)

disks attached to MicroVAX and VAX processors.

This chapter discusses:
o

Automatic bad block replacement —

how the ULTRIX-32 operating

system does dynamic bad block replacement
o

Disk error messages interpretation-

how the system detects, records,

and reports errors
e

Forced error recovery—

how to recover the data associated with a bad

block
o

The radisk utility -

how to replace, clear, and scan blocks on a disk

manually

4.1

Automatic Bad Block Replacement

There are three strategies to replace bad blocks for DSA disks:
initiated, host-initiated, and media replacement.

controller-

The controller-initiated and

host-initiated strategies involve replacing a bad block with a good block
reserved for this purpose by the DSA disk subsystem.
involves replacing the medium.

The third strategy

Device support for these three strategies is

summarized in Table 4-1.

4.1.1

Controller-Initiated and Host-Initiated Strategies

The controller-initiated and the host-initiated bad block replacement
strategies have the same result, although they are implemented differently.
In the controller-initiated strategy, the hardware confirms and replaces a

detected bad block.

In the host-initiated strategy, the controller notifies

the host software of a bad block. Then the host software confirms and
replaces the bad block.

Table 4-1: Device

Strategy

Support for Bad Block Strategies

Controller(s)

Controller-initiated

=~ RQDX1, RQDX2, RQDX3

Device( s)

RD31, RD32, RD51, RD52,
RD53, RD54

Host-initiated

Media replacement

Small VAX disk controller*

RD31, RD32, RD53, RD54

UDA50A, KDA50, KDB50

RA60, RA80, RA81, RA82

HSC50, HSC70

RA70, RA90

RQDX1, RQDX2, RQDX3

RX50, RX33

RUX50

RX50

small VAX disk controller

RX50, RX33

*The small VAX disk controller is used by the VAXstation 2000 and MicroVAX
2000 processors.

The software driver for the small VAX disk controller emulates

RQDX3-style controller-initiated bad block replacement.

4.1.2

The

Media

Replacement Strategy

The RX50 and RX33 devices have no means of replacing bad blocks.

If a

hard error with a bad block is reported, the diskette must be replaced.
1.

First try the diskette in another drive to be sure the problem is with

the diskette and not the drive.
2.

If you can read the diskette, make a copy of it, since the original
diskette may be marginal.

File the original and use the copy.

If the diskette is bad, recover as much of the data as you can, and
recreate the remainder of the diskette.

4.2

Disk Error Messages

Interpretation

This section describes how the system detects and reports disk errors and
how to interpret bad block error messages.

4-2

Bad

Block Replacement

4.2.1

Logical

Block Numbers

The DIGITAL Storage Architecture (DSA)

does not use the actual physical

address of sectors on the medium. Instead, a sector on the medium is
addressed by a logical block number (LBN).

An LBN is the basic

addressable unit of the user-accessible data area of the disk. The first LBN
on a disk is numbered 0.

The highest numbered LBN on a disk is

numbered one less than the number of LBNs in the user-accessible area of
the disk.

The LBN reported in an error message is relative to the start of the
physical disk unit and is that of the first bad block encountered.

4.2.2

Detecting and

Reporting Disk Errors

The disk controller detects and reports data errors when bad blocks cause
erroneous data to be transferred from the drive to the controller.

Correction Code (ECC)
erroneous data found.

Error

schemes are used to detect and correct any
These schemes apply statistical algorithms to

dynamically correct erroneous data and to successfully complete requested
I/O operations.

If the data from a bad block is not correctable, the

controller performs a series of rereads and other recovery techniques and
attempts to obtain the data and deliver it without error.

Using ECC schemes, the hardware can correct an error when it is within a
certain limit.

If the error is equal to or below the limit, the controller can
correct it and complete the I/O operation successfully without a retry.
If

the error is above the limit, the controller cannot correct the data.

After

repeated retries, the I/O operation fails and the failure is reported as a
hard error to the host operating system.
The block is bad and needs to

be replaced.
Errors that are correctable but that are above a given threshold are also
reported to the host operating system. These marginal blocks are replaced
to prevent future loss of data.

If the data read from the bad block resulted in an uncorrectable ECC
error, the replacement block is written with the forced error indicator. A
forced error indicator warns that the data may be corrupted.

4.2.3

Examining the

Errorlog File

The messages in the kernel errorlog buffer are logged to the errorlog file
on disk, where they are accessed by the uerf command. The default
errorlog file is /usr/adm/syserr/syserr.hostname.

To produce an error report containing driver error messages from the
errorlog file, run the uerf command with the —r option:
#

/etc/uerf

-r

250

Bad

Block Replacement 4-3

This command accesses ASCII messages. Driver error messages are logged
in ASCII rather than binary format.

The -r option does not produce error messages for the VAXstation 2000 or
MicroVAX 2000 disk driver. To produce an error report with messages for
the sdc driver, use this command:
#

The ?

/etc/uerf

-c

oper

grep

sd?

should be the unit number.

To diagnose a problem, study the types of errors logged.
For example, a
sudden incidence of hard errors reported at random places on a device may
indicate an electrical or mechanical problem.

For more information on uerf(8) see the ULTRIX Reference Pages and
Guide to the Error Logger System.
Note

Defective hardware can cause good blocks to be reported as bad.
The system may be replacing blocks that would not be considered
bad if the hardware was functioning properly. Check your
hardware first if an error report displays an unusual number of
bad blocks.

4.2.4

Interpreting Error Messages

The uda and sdc device drivers report bad block information to the
errorlog file.
Several types of error messages can be accessed from this
file.

The following sections describe the disk error messages that may appear in
the errorlog file. The examples are in the terse output format.
Note

The text of an ASCII error message, when reported in the brief
or full output format, will generate more than one entry in the
errorlog file. These ASCII text messages may not print
sequentially.

4.2.4.1

Bad Block Reported Message - When an error results from a

bad block reported condition, this message appears on the error report:
sd0:

Bad
LBN

4-4 Bad

Block
79

Reported

replaced

Block Replacement

LBN

This message identifies the device and the location of the LBN and
indicates that the block is bad and has been replaced.

4.2.4.2

Bad

Blocks Not Reported

Message

When the controller

encounters more than one bad block during a read of multiple blocks, it

returns the LBN of the first block only. The following message appears on

the error report:
udab50a0:

unit

Bad

unit

Bad

Block

Blocks

LBN

Reported

NOT

42345

LBN

42345

Reported!

replaced

This message indicates that the reported block is bad and has been
replaced.

It warns that there is at least one more bad block.

A similar

message reports the next bad block and replaces it.

Note

The VAXstation 2000 and MicroVAX 2000 processors do not

produce this message.

4.2.4.3

Transient Error Messages -

Transient errors are caused by

malfunctioning hardware or by marginal media.

If your hardware is

properly functioning, transient errors recorded in the errorlog file repeatedly
for one LBN may indicate an impending bad block.
To check for transient errors, use the uerf command to produce an error
report.

Here is an example of a transient error message:

kdb501:

unit

Bad

unit

Transient

Block

Reported

error

LBN

179368

179368,

Block

not

bad

This message indicates that the reported block is not bad and has not

been replaced.
When transient errors occur repeatedly, the controller’s recovery mechanism
may eventually fail to read the data successfully. If the block on the

medium degrades, and the error level becomes too high to be corrected by
ECC schemes, the transient error may become a hard error.

4.2.4.4

Forced

Error Condition -

During a bad block replacement

operation, data is read from the bad block and written to the replacement

block.

When valid data cannot be read from the bad block and written to

the replacement block, the forced error condition is set for the replacement

Bad

Block Replacement 4-5

block to indicate corrupted data.

This message occurs when the hardware

reads a block written with the forced indicator set:
rala:

hard

rala:

Force

error

Error

123

Modifier

set

LBN

12345

The first line of this example appears at the console and the entire
message is logged to the errorlog file.

This message means that the data in the block may be corrupted and
should be restored. Restoring the data causes the block to be rewritten;
this clears the forced error condition because the data is restored to a
correct state.

4.3

Forced

Error Recovery

When a read of a bad block cannot be successfully completed, the

controller sets the forced error indicator in the replacement block.

This

indicator warns that the data written in the replacement block may be

corrupted. If a read is successfully completed, the forced error indicator is
not set and the replaced block is considered clean.

4.3.1

Discovering Forced

Errors

To find out if there are any forced error conditions on a disk, run the uerf
command with the -r option:
#

/etc/uerf

-r

250

This command accesses ASCII messages. Driver error messages are logged
in ASCII rather than binary format.

The -r option does not produce error messages for the VAXstation 2000 or
MicroVAX 2000 processors. To produce an error report with messages for
the sdc driver, use this command:
#

/etc/uerf

-c

oper

grep

sd?

The ? should be the unit number.

4.3.2

Restoring Data

When you discover a forced error indicator on a disk, you may not know
what type of data is associated with the block generating the error.
Before you restore a block reporting a forced error, you should determine
what kind of information the block contains:

Shut down multiuser mode.
system to single-user mode.

4-6

Bad

Block Replacement

Use the shutdown command to bring the

Use the icheck command with the —b option to discover how the
block is used.

You need the sector number of the bad block from the errorlog
message and the specification of the partition within which the bad
:

block occurred; do not use the LBN.
#

icheck

-b

2300

/dev/rraOg

If the icheck command produces an inode number, indicating the
block is part of a file, specify the ncheck command with the -i

option to determine the corresponding file. Use the inode number
and the partition specification:
#

ncheck

-i

354

/dev/rra0Og

Output from this command shows the file name or names associated
with the bad block.

If the block with the forced error is a data file, copy the data to a
new file and delete the old corrupted file. If the data is still corrupt,
restore the file from a backup. Then use the fsck command to clean
the file system. If the errors are still not cleared, run fsck again.
You may need to run fsck several times.
If the forced error is in the superblock, use the backup superblock

(fsck —b). See the fsck command for more information.
If the forced error is in an inode or a cylinder group block, first try
the fsck command.

If fsck does not work, decide whether to restore

the file system or clear the block. If you decide to clear the block,
use the radisk utility.

The

radisk Utility

The radisk utility allows you to replace, clear, and scan blocks on a disk
manually. However, you must have superuser privileges, be in single-user

mode, and have all the file systems except root unmounted when using
this command.

The radisk command has three options:
—C clear
—r replace
—$S

scan

Bad

Block Replacement 4-7

Note

The VAXstation 2000 and MicroVAX 2000 processors do not
support the -r option of the radisk command.

The format for the radisk command is:
radisk —option

LBN length special

LBN is the number of the logical block. The length is the number or

blocks from the specified LBN that you want to scan or clear.
Special
refers to the raw device special file ( /dev/rra3c, for example). For the -c¢
and -r options, the special file specified indicates an unmounted ¢ partition

of a character device special file. The —s option will accept any valid
partition on the disk.

The system produces an error message with both an

LBN and a sector number.

4.4.1

The

Clear Option

The clear option clears the forced error indicator in the specified area of a
disk. The LBN can then be read without generating an error, even though

the block may contain corrupted data.

This command line clears the forced error indicator from LBN 12334 to
the last block of the partition:
#

radisk

-c

12334

-1

/dev/rra3c

Use -1 to refer to the end of a disk partition. In this example, -1
represents the number of LBNs from LBN 12334 to the end of the disk.

Note

Clearing the block will remove the forced error indicator.
However, the data in the block must still be considered
corrupted. To maintain data integrity, check the data, and if
necessary, restore it from backup. See the Guide to System

Backup and Restore for information on restoring data.

4.4.2

The

Replace

Option

The -r option forces the replacement of blocks on a disk.

This option replaces a single block on a DSA disk. It enables you to
replace a block that records repeated transient errors before it goes bad,
thus saving the data.

4-8

Bad

Block Replacement

For example, this command line forces LBN 12345 to be replaced:
#

radisk

-r

12345

/dev/rra3c

This option applies to host-initiated bad block replacement only; it does not
apply to controller-initiated replacement.
Note

Replacing blocks affects performance. As blocks are replaced, the
speed of the system is affected. In addition, there are a limited
number of replacement blocks available on a disk.

4.4.3

The Scan Option

The -s option scans a disk for bad blocks. Use this option after
reformatting or when the disk is new. The scan option reads the specified
range of blocks and reports any errors. If a bad block is found, the block
is replaced and the scan option restarts and rescans the specified range
until it can complete without reporting any bad blocks.

When a forced error condition is found, radisk reports the LBN and
continues to scan. It does not restart and the forced error indicator is not
cleared. Use the —c option to clear a forced error condition reported during
a

scan.

;

This command line scans the g partition starting at LBN 12345 for 2000
blocks:
#

radisk

-s

12345

2000

/dev/rra0Og

For the —s option, an LBN of 0 means the first LBN of the special file’s
partition.

‘

For more information on the scan option, see the radisk(8) command in
the ULTRIX Reference Pages.

Bad Block Replacement 4-9

Device Mnemonics A

This appendix identifies and defines the mnemonics that are used to attach

any hardware or software device to your system.

The mnemonics are used

by the /deviIMAKEDEV shell script to create the character or block special
files that represent each of the devices.

The mnemonics also appear in

the system configuration file as described in the Guide to System
Configuration File Maintenance.

Table A-1 lists the mnemonics in seven categories: generic, consoles, disks,
tapes, terminals, modems, and printers.

The generic category lists the

mnemonics of a general nature and includes memory, null, trace, and tty

devices.

The consoles category lists the system console devices that the

ULTRIX operating system uses.

The disks, tapes, terminals, modems, and

printers categories identify the appropriate mnemonics for those devices.

The description heading in Table A-1 identifies the corresponding device
name.

It does not define the mnemonic’s use.

For detailed information

on the use of each mnemonic in relation to both the MAKEDEV script and
the system configuration file, refer to the reference pages in Section 4 of

the ULTRIX Reference Pages.

If on-line reference pages are available, you

can also use the man command.

For instance, if you enter at the system

prompt:
#

man

the system displays the reference page for the Mass Storage Control
Protocol (MSCP)

disk controller driver.

Where appropriate, the SYNTAX

section of the reference page defines the device’s syntax as it appears, or
should appear, in the config file. Refer to /dev/IMAKEDEV for additional
software device mnemonics that MAKEDEV uses.

Refer to MAKEDEV(8)

the ULTRIX Reference Pages for a description of the MAKEDEV utility.
You should note that Table A-1 uses the convention of an asterisk (*)

beside a mnemonic and a question mark (?)

beside a device name to mean

a variable number. The range of the variable number is dependent on the
particular device.

Table A-1: Devices Supported by MAKEDEV