Digital PDFs

AA-ME93A-TE

May 1988

64 pages

Original

2.1MB

Document:	ULTRIX-32 Guide to System Disk Maintenance
Order Number:	AA-ME93A-TE
Revision:	000
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
DEC US
MASS BUS
MicroVAX
PDP

Q-bus
RT
ULTRIX
ULTRIX-11
ULTRIX-32
UNIBUS

VAX
VAXstation
VMS
VT
ULTRIX Worksystem Software
~amaomo

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 Ma nua I
Audience ......... .. .. ............... ................................. .. ................................ ....... ....

vii

Organization ....................................................................................................

vii

Related Documents ................................... .. .............. ............... ........... .... ... ....

viii

Conventions ........................................... .. ... .. .... .. ........ ........... .... ......... .............

viii

1 System Disks
1.1 Disk Organization

1-1

1.1.1 Disk Partition Defaults ..............................................................
1.1.2 Selecting Disk Partitions ...........................................................

1-2
1-2

1.2 Paging and Swapping .............................................................................

1-2

1.2.1 Allocating Swap Space ...............................................................

1-3

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

1-4

1.3.1 Setting Up /var ...........................................................................

1-4

1.4 System Directories ..................................................................................

1-6

1.5 User Directories ......................................................................................

1-7

1.5.1 Creating a User Directory Tree Structure ..............................
1.5.2 Sample User Directory Tree Structure ....................................

1-7
1-8

1.5.2.1 Clerical Users ...........................................................................

1-9

1.5.2.2 Scientific Users ........................................................................ 1-10
1.5.2.3 Administrative Users ............................................................... 1-10

2 Managing Disk Space
2.1 Monitoring File System Use .................................................................

2-1

2.1.1 Checking Available Free Space .................................................
2.1.2 Checking Disk Use .....................................................................
2.1.3 Verifying Disk Quotas ................................................................

2-1
2-2
2-2

2.2 Obtaining Additional Disk Space ..........................................................

2-3

2.2.1 Deleting Unused or Obsolete Files ...........................................
2.2.2 Moving a File System ................................................................
2.2.3 Moving Files to a Larger File System ....................................

2-3
2-3
2-4

3 Disk Partitioning
3.1 Preparing to Change Disk Partitions ...................................................

3-1

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

3-2

3.2.1 Partition Table Values ................................................................
3.2.2 Reading the Partition Tables .....................................................
3.2.3 Rules for Changing Partition Tables ........................................

3-2
3-3
3-3

3.3 Reviewing the Current File System Assignments ..............................

3-5

3.4 Reviewing the Current Disk Partition Setup .....................................

3-5

3.5 Changing Partition S:izes ........................................................................

3-6

3.5.1 Changing Partition S:izes on an RA81 Disk ...........................
3.5.2 Changing Partition S:izes on an RM05 Disk (Example) ......

3-7
3-9

4 Bad Block Replacement
4.1 Automatic Bad Block Replacement
4.1.1 Controller-Initiated and Host-Initiated Strategies ...................

iv Contents

4-1
4-1

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

4-2

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

4-2

4.2.1 Logical Block Numbers ..............................................................
4.2.2 Detecting and Reporting Disk Errors ......................................
4.2.3 Examining the Errorlog File .....................................................
4.2.4 Interpreting Error Messages ......................................................

4-3
4-3
4-3
4-4

4.2.4.1 Bad Block Reported Message .................................................
4.2.4.2 More Bad Blocks Not Reported Message ............................
4.2.4.3 Transient Error Messages .......................................................
4.2.4.4 Forced Error Condition ...........................................................

4-4
4-5
4-5
4-5

4.3 Forced Error Recovery ...........................................................................

4-6

4.3.1 Discovering Forced Errors .........................................................
4.3.2 Restoring Data ............................................................................

4-6
4-6

4.4 The radisk Utility ...................................................................................

4-7

4.4.1 The Clear Option ........................................................................
4.4.2 The Replace Option ....................................................................
4.4.3 The Scan Option .........................................................................

4-8
4-8
4-9

A Device Mnemonics
Index

Figures
1-1:

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

1-5

1-2:

Subdirectories of /usr, One-Branch Tree ..........................................

1-8

1-3:

Subdirectories of /usr, Three-Branch Tree .......................................

1-9

Contents v

Tables
4-1:

Device Support for Bad Block Strategies ........................................

4-2

A-1:

Devices Supported by MAKEDEV ...................................................

A-2

vi Contents

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. It
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.

Related 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. For example: See the
cat(1) command. 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.

fun ct i on

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.

>>>

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.
<KEYNAME > In examples, a word or abbreviation in angle brackets
indicates that you must press the named key on the
terminal keyboard.
<CTRL/x>

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

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:

•
•
•
•
•
•
•
•
•
1.1

Disk organization
Disk partition defaults
Selecting disk partitions
Paging and swapping
Allocating swap space
File system organization
Setting up /var
System directories
User directories

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 O) 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.

•

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. The hardware is designed to process programs in lengths
called pages. 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. This writing and reading of modified pages of memory is

1-2 System Disks

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) m
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. If
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. You do not
usually need to change the partitions of the disk. 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 Net work 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. Variable data files reside in /var or /usr/var. On server
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.
Executable data files are machine architecture-specific files. All but a
select set of executables can be shared by most systems. 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. Separating files by
content simplifies the system management task. 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, I
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.

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.

•

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. If
you installed your system with the Advanced Installation, you were
provided with the option of creating a separate file system for
/usr/users. 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) directory for users is subordinate to
/usr/users. 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.
•

•
•

1.5.1

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.

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

l
/users

]
I

/o.cla.MS

/Jones

/SMlth

/'
/za.blonsky

ZK-001 SU-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

/clerlco.l

/JIM

/Ma.ry

/O.dMln

I scientific

/fred

/ha.rry

la.lice

/sa.ro.h

/don

/Jeff

B
ZK-0016U-GE

Figure 1-3: Subdirectories of /usr, Three-Branch Tree
1.5.2.1 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. We will assume that these files require an average of
30 Kbytes each.

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 e on an
RA81 disk, for example).
1.5.2.3 Administrative Users - The administrative staffs needs are also
fairly large, but they have a special security requirement. They keep
salary and personnel files that must remain private. 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. Refer to newfs( 8)
in the ULTRIX Reference Pages. 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. By
using partition g, the file system will have more room than if you use
partition d, e, or f. To review the current disk partition setup, use the
chpt program. See Chapter 3 of this manual and the ULTRIX Reference
Pages for details on the chpt command.

1-10 System Disks

Managing Disk Space

This chapter discusses methods of monitoring file system use and managing
file system data. It explains how to check the disk space and use and
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) using quot

2.1.1

Checking Available 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:
# df

Fi I esystem
node
/dev/raOa
/dev/rala
/dev/ralg
/dev/raOe
/dev/raOh

total
kbytes
7429
7429
175015
30519
313233

kbytes kbytes
used
free
6141
546
437
6250
143520
13994
14342
13126
134316 147594

percent
used
923
73
913
523
483

Mounted on
I

/tmp
/usr
/usr/spool
/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. For more information on the df command and
its options, see df( 1) 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 newts
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) m the /usr/users file system:
# du -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. For more information on checking
disk space use, refer to quot( 8) and du( 1) 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.
user.

This command allows you to set or change the limits for each
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
Supp"lementary Documents, as well edquota( 8), quotacheck( 8), quotaon( 8),
quotaoff( 8), and repquota( 8) in the ULTRIX Reference Pages.

2.2

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:
•

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.
2.2.1

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:
1.

Dump the file system (level 0) with the dump command.

2.
3.

Create a clean file system with the newts 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: moving all
the files and merging individual files. 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. To create and
restore individual files, use the tar command. For example, to create the
file named file1, type:
#

tar c f i I e 1

In this example, the tar command creates file1 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) the file from the tape
device to the target file system. For example,
#tar xp fi lel

This command extracts file1 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:
•

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.

Preparing to Change Disk Partitions

3.1

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. Here you might ask:
•

Does the current disk partitioning setup optimize the average
and peak demands of system users?

•

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:
•

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.

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 RM05
disks on the system with different partitions, one using the default values
for all RM05 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 c partition does not have a file system for a partition table
to reside in.

An individual partition table was never created for the disk.

3.2.3

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 newts 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.

tracks/cyl)

(No. sect/cy I)

(No. sect/cyl) * (cyl No.) =sect

For example, suppose you have an RM05 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 rm OS
rm05: #sectors/tracks=51, #t racks/cy Ii nde rs=14
#cy Ii nders=1248
partition
size
range
15884
0 - 26
a
33440
27 - 81
b
500384
0 - 822
c
d
15884
562 - 588
55936
589 - 680
e
f
86176
681 - 825
g
158528
562 - 822
291346
h
82 - 561

Now you can perform the calculation:
51 * 14

714

714 * 5

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:
# df
Filesystem
node
/dev/raOa
/dev/raOh

total
kbytes
7429
388950

kbytes
used
5371
118000

kbytes
free
1315
270950

percent
used
803
303

Mounted on
I

/usr

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 ch pt 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 c. For example, to check current partition
sizes on an RA81, type:
# chpt

q /dev/rraOa

Disk Partitioning 3-5

The chpt program responds with output similar to this:
Current partition table:
top
overlap
partition
bottom
size
15883
15884
c
a
0
15884
49323
33440
c,h
b
891071
a,b,d,e,f,g,h
c
891072
0
h,c
d
340670
356553
15884
412489
h,c
e
356554
55936
f
412490
891071
478582
h,c
g
49324
131403
82080
c
h
131404
891071
c,d,e,f
759668

3.5

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:
/etc/chpt - a -q -d -v - px offset size ... device

For more information on changing partition sizes, see chpt( 8) m the
ULTRIX Reference Pages.
The following sections explain the procedure and give examples for:
•

Changj.ng partition sizes on an RA81 disk

•

Changing partition sizes on an RM05 disk

3-6 Disk Partitioning

3.5.1

Changing Partition Sizes on an RA81 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 8ystem 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
partition table:
#

chpt

q option to see the status of the

q /dev/rra7a

/dev/rra7a
Current partition table:
partition
top
bottom
0
a
15883
b
15884
49323
0
891071
c
d
340670
356553
356554
412489
e
f
412490
891071
g
49324
131403
h
131404
891071

size
15884
33440
891072
15884
55936
478582
82080
759668

overlap
c
c,h
a,b,d,e,f,g,h
c,h
c,h
c,h
c
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 on /usr
hpOh on /usr/staff
ra7a on /mnt

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. Here is the formula that
you can use:
(new length of a)= (beginning of b) +(size of b)

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 - Here is a correct way to extend the length of
an active partition. This example shows how to extend the a
partition by the size of partition b:
# chpt -v -pa 0 49324 /dev/rra7a
/dev/rra7a
Current partition table:
partition
top
bottom
size
a
0
49323
49324
b
15884
49323
33440
c
0
891071
891072
d
340670
15884
356553
e
356554
412489
55936
f
412490
891071
478582
g
49324
131403
82080
h
131404
891071
759668

3-8 Disk Partitioning

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

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:
# chpt -pa 0 800 /dev/rra7a
chpt: cannot set partition table
Mount device busy

in driver:

Method C - Incorrect - Another rule is that you can never change
the starting sector (offset) of partition a:
# chpt -pa 6 800 /dev/rra7a
chpt: cannot set partition table
Mount device busy

in driver:

Step 6: Make and Mount File Systems

Then, place the file system on the partition, using the newfs command.
# /etc/newfs /dev/rra7a ra81
Warning: partition table overriding /etc/disktab
Warning: 656 sector(s) in last cylinder unallocated
/dev/rra7a:
49324 sectors in 70 cylinders of 14 tracks,
51 sectors 25.3Mb in 5 cyl groups
(16 c/g, 5.85Mb/g, 2048 1/g)
superblock backups (for fsck -b#) at:
32, 11520, 23008, 34496, 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.
use the restore command.
Step 8:

If there were, you would

Check File Consistency

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

3.5.2

Changing Partition Sizes on an AMOS Disk (Example)

Although this example is for an RM05 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 RM05 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. Therefore, you need to double
the size of partition b. 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 RM05 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 RM05 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 RM05 disk, there are 32 sectors per track and 19 tracks. Thus
there are 608 sectors per cylinder ( 32*19). There are 823 cylinders.
Here is the default table, found in hp( 4) for an RM05 on drive 0:
Disk

Start

Length

Cylinders

hpOa
hp Ob
hpOc
hpOd
hpOe
hp Of
hpOg
hp Oh

0
16416
0
341696
358112
414048
341696
49856

15884
33440
500384
15884
55936
86240
158592
291346

0-26
27-81
0-822
562-588
589-680
681-822
562-822
82-561

Here is how you want the partitions to be divided.
partitions being modified are highlighted:

The lengths of the

Disk

Start

End

Length

Cylinders

hpOa
hp Ob
hpOc
hpOd
hpOe
hp Of
hpOg
hpOh

0
16416
0
341696
358112
414048
341696
83296

15884
83295
500384
357579
414047
500287
500287
341202

15884
66880*
500384
15884
55936
86240
158592
257906*

0-26
27-136
0-822
562-588
589-680
681-822
562-822
137-561

3-10 Disk Partitioning

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
q /dev/rhpOa
/dev/rhpOa
No partition table found in superblock ...
using default table from device driver.
Current partition table:
overlap
partition
bottom
top
size
15884
a
O
15883
c
b
16416
49855
33440
c,h
500384 a,b,d,e,f,g,h
c
0
500383
d
341696
357579
15884
c,g
e
358112
414047
55936
c,g
c,g
f
414048
50028
86240
g
341696
500287
158592 c,d,e,f
h
49856
341201
291346 c

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 RM05 disk on drive 0.
Before you can change the partition table, you must copy the RM05 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 RM05 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 RM05 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 of b) = (current size of b) x 2
(beginning of h) = (size of new b) + (beginning size of b)
(size of h) = (ending sector of h) - (beginning sector of h +1)

The calculation for this example is:
66880
83296
257906

33440 x 2
66880 + 16416
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/rhpOa
New partition table:
partition
top
bottom
size
a
0
15883
15884
b
16416
83295
66880
0
500383 500384
c
d
341696
357579 15884
e
3 58112
414047 55936
f
414048
500287 86240
g
341696
500287 158592
h
49856
341201 291346

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

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

3-12 Disk Partitioning

# chpt -v -ph 83296 257906 /dev/rhpOa

/dev/rhpOa
New partition
partition
a
b
c
d
e
f
g
h

table:
bottom
0
16416
0
341696
358112
414048
341696
83296

top
15883
83295
500383
357579
414047
500287
500287
341201

size
15884
66880
500384
15884
55936
86240
158592
257906

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

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

rmOS

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 Micro VAX and VAX processors.
This chapter discusses:
•

Automatic bad block replacement - how the ULTRIX-32 operating
system does dynamic bad block replacement

•

Disk error messages interpretation- how the system detects, records,
and reports errors

•

Forced error recovery- how to recover the data associated with a bad
block

•

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: controllerinitiated, host-initiated, and media replacement. 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. The third strategy
involves replacing the medium. 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 Support for Bad Block Strategies
Strategy

Controller( s)

Device( s)

Controller-initiated

RQDXl, RQDX2, RQDX3

RD31, RD32, RD51, RD52,
RD53, RD54

Small VAX disk controller*

RD31, RD32, RD53, RD54

UDA50A, KDA50, KDB50

RA60, RASO, RA81, RA82

HSC50, HSC70

RA70, RA90

RQDXl, RQDX2, RQDX3

RX50, RX33

RUX50

RX50

small VAX disk controller

RX50, RX33

Host-initiated

Media replacement

*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.

If you can read the diskette, make a copy of it, since the original

diskette may be marginal.
3.

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 LBN s 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. Error
Correction Code ( ECC) schemes are used to detect and correct any
erroneous data found. 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 lusr/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
Micro VAX 2000 disk driver. To produce an error report with messages for
the sdc driver, use this command:
# /etc/uerf

The ?

-c oper I 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:
sdO: Bad Block Reported at LBN 79
LBN 79 replaced

4-4 Bad Block Replacement

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 More 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:
uda50a0:

unit 1, Bad Block Reported at
More Bad Blocks NOT Reported!
unit 1, LBN 42345 rep I aced

LBN 42345

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 Micro VAX 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 5,
unit 5,

Bad Block Reported at LBN 179368
Transient error on LBN 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:
rala:

hard error sn 123
Force Error 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
Micro VAX 2000 processors. To produce an error report with messages for
the sdc driver, use this command:
# /etc/uerf -c oper

I 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:
1.

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

5.
6.

4.4

354 /dev/rraOg

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 Micro VAX 2000 processors do not
support the - r option of the radisk command.
The format for the radisk command is:
radisk - option

LBN kngth 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 c 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 LBN s 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 Sean 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/rraOg

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 /dev/MAKEDEV 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/MAKEDEV for additional
software device mnemonics that MAKEDEV uses. Refer to MAKEDEV( 8) in
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
Category

Mnemonic

Description

Generic

boot*
mvax*
vaxstation *
std
drum
err log
kUmem
kmem
mem
null
trace
tty
local

Boot and std devices by cpu number; e.g., boot750
All Micro VAX setups; e.g., mvax2000
A VAXstation 2000 setup; e.g., vaxstation2000
Standard devices below with all console subsystems:
Kernel drum device
Error log device
Kernel U nibus/Q-bus virtual memory
Virtual main memory
Physical memory
A null device
A trace device
A tty device
Customer specific devices

Consoles

console
crl
cs*
ctu*
cty*
cfl
ttycp

System console interface
Console RL02 disk interface for VAX 86?0
Console RX50 floppy interface for VAX 8??0
Console TU58 cassette interface for VAX 11/7 50
Console extra serial line units for VAX 8??0
Console RXOl floppy interface for 11/78?
Console line used as auxiliary terminal port

Disks

hp*
ra*
ese*
rb*

MASSBUS disk interface for RM?? drives
UNIBUS/Q-bus/BI/HSC MSCP disk controller interface
UNIBUS/Q-bus/BI/HSC MSCP electronic ESE20 disk
UNIBUS IDC RL02 disk controller interface
for RB?? drives
VAXstation 2000 and Micro VAX 2000 RD type drives
SCSI disks ( RZ22/RZ23/RZ55/RRD40)
UNIBUS RK?? disk controller interface
UNIB US/Q-bus RL ?? disk controller interface
VAXstation 2000 and MicroVAX 2000 RX type drives

rd*

rz
rk*
rl*
rx*
Tapes

mu*
tms*
rv*
ts*
tu*
st*

A-2 Device Mnemonics

TU7 8 MASS BUS magtape interface
UNIBUS/Q-bus/BI/HSC TMSCP tape controller interface
UNIBUS/Q-bus/BI/HSC TMSCP optical disk
UNIBUS/Q-bus TS11/TS05/TU80 magtape interface
TE16/TU45/TU77 MASSBUS magtape interface
VAXstation 2000 and Micro VAX 2000 TZK50
cartridge tape