Professional Documents
Culture Documents
AIX Manual sysmgt
AIX Manual sysmgt
System Management
Table 5 (Page 1 of 2). AIX Version 3.2/AIX Version 4.1 Printer Name Changes
AIX Version 3.2 AIX Version 4.1 AIX Version 3.2 AIX Version 4.1
name name name name
2380 ibm2380 2380-2 ibm2380-2
2381 ibm2381 2381-2 ibm2381-2
2390 ibm2390 2390-2 ibm2390-2
2391 ibm2391 2391-2 ibm2391-2
3812-2 ibm3812-2 3816 ibm3816
4019 ibm4019 4029 ibm4029
4037 ibm4037 4039 ibm4039
4070 ibm4070 4072 ibm4072
4076 ibm4076 4079 ibm4079
4201-2 ibm4201-2 4201-3 ibm4201-3
4202-2 ibm4202-2 4202-3 ibm4202-3
4207-2 ibm4207-2 4208-2 ibm4208-2
4212 ibm4212 4216-31 ibm4216-31
4224 ibm4224 4226 ibm4226
4234 ibm4234 5202 ibm5202
5204 ibm5204 5575 ibm5575
Also, removing the AIX Version 3.1 symbolic links (4.7.6, “AIX Version 3.2
Symbolic Links Being Repackaged” on page 23) may affect users who do not
install the symbolic link compatibility package. For example, the /usr/lpd/pio/ddi
files are actually in the /var/spool/lpd/pio/@local directory.
See 4.7.3, “Printer Subsystem Packaging” on page 22 for other printer packaging
information.
Volume group images will be saved in backup format (BFF). In AIX Version 3.2,
the rootvg image was saved in tar format. The rootvg is created as an
installable image; other volume groups may be backed up as separate images
on separate tapes.
If the mksysb tape is used for a backup of the source system, it is considered a
system backup . However, if the intent of the backup is to provide a customized
system for use on other machines, the mksysb is considered a clone. Cloning
means preserving either all or some of a system′s customized information for
use on a different machine.
In AIX Version 4.1, the running system will be ″custom″ tailored to the hardware
on that system; that is, only devices installed on the system will be supported. If
the user needs additional device support for other systems, it will have to be
specifically installed onto the running system before the mksysb is created.
7.2.1.3 Commands
1. mksysb - Create an installable image of the rootvg onto a bootable tape
media or to a file. The mksysb command will only backup the rootvg volume
group.
The Type 6015 Model 0U0 PowerPC Personal System does not support
booting from a tape or making a boot image on a tape. If the system is a
Type 6015 Model 0U0 PowerPC Personal System, the tape will not have a
″boot image″; it will only have a bos Install image and then the actual
mksysb image. In other words, the tape format will be:
RISC System/6000 system - boot image,
bos install image, dummy TOC, mksysb image
Type 6015 Model 0U0 PowerPC Personal System - dummy boot image,
bos install image, dummy TOC, mksysb image
7.3.2 GUI Support for Devices, LVM, Printers, and Users/Groups (AIX Version
3.2.5 Work)
Provide a graphical user interface primarily aimed at novice users. There will
be icons to support the following system management activities:
• Management of system access, including password
• Management of logical volumes, volume groups and physical volumes
• Management of printers and their queues
• Management of device configuration
The flags:
-s When this flag is specified, the ypbind daemon is run in secure mode.
The daemon will use only priviliged ports for communication.
-ypset If this flag is specified, the local host will accept ypset commands
from the local host and remote hosts.
-ypsetme If this flag is specified, the local host will accept ypset commands
from only the local host. Any ypset commands from remote hosts will
be rejected. This flag will override the -ypset flag if both are
specified.
Note: If neither the -ypset or -ypsetme flags are specified, the local host will
reject all ypset commands from all hosts. This is the most secure mode since
the NIS server can′t be changed.
Fragment support applies to the last partial direct block of small user files and
directories, and long symbolic links. Fragment size is specified for a file system
at creation time and can be a power of 2 division of the file system full block
size, with a maximum fragment factor of 8. In addition, different file systems
may have different fragment sizes, and they may coexist simultaneously within a
single system. Given the full block size of 4096 bytes, the allowable fragment
sizes for JFS file systems are 512, 1024, 2048, and 4096 bytes. For consistency
with AIX Version 3, the default fragment size is 4096 bytes.
Variable number of disk inode support allows users to specify the number of disk
inodes to be created within a file system if more or less than the default number
of disk inodes are desired. The number of desired disk inodes may be specified
at file system creation as the number of bytes per inode (nbpi). For example, a
nbpi value of 1024 causes a disk inode to be created for every 1024 bytes of a
file system. For JFS file systems, allowable nbpi values must be a power of 2
multiple of the minimum value of 512 and can be as large as 16384. For
consistency with AIX Version 3, the default nbpi value is 4096.
The implementation of the fragment and variable number of inode support will
provide full AIX Version 3 disk image compatibility for file systems and file
system logs, allowing AIX Version 3 file systems to be mounted and accessed in
the usual manner without requiring customers to perform any migration
activities. In addition, the provided disk image compatibility will allow AIX
Version 3 file systems to be interchanged between AIX Version 4.1 and AIX
Version 3 systems. For example, an AIX Version 3 file system can be mounted
Under AIX Version 4.1, AIX Version 3 file systems will be created when both the
file system fragment size and nbpi values are equal to AIX Version 3 values of
4096 bytes. This is done to allow customers to create file systems under AIX
Version 4.1 that can be interchanged between AIX Version 4.1 and AIX Version 3
systems. Since the AIX Version 4.1 default value for both fragment size and nbpi
is 4096 bytes, AIX Version 3 file systems will be created by default.
An AIX Version 4.1 file system will be created when the specified fragment size
or nbpi values are other than 4096 bytes. These file systems will not be
compatible with AIX Version 3 systems; this older version of AIX will have little
hope of understanding the new JFS file system format, including fragment
allocation units, allocation map search tree encodings, the format of file system
disk addresses, and variable numbers of disk inodes. In fact, AIX Version 4.1 file
systems will have a new superblock magic number that will prevent these file
systems from being recognized by the AIX Version 3 implementation. To move
an AIX Version 4.1 file system to an AIX Version 3 system, significant migration
activities (mkfs, dump, restore) will be required. Similar activities will have to be
performed on AIX Version 4.1 systems if fragment or nbpi support is desired for
an existing AIX Version 3 file system.
File compression helps reduce the price/MB of disks and thus helps reduce the
overall price of the system.
Compression only applies to regular files and long symbolic links; normal
fragment support is used for directories and other metadata is not compressed.
Each logical block of a file is compressed by itself. Normally, the size of the
compressed block is less than a full block, in which case it is written in
compressed form on disk in only the number of contiguous fragments needed.
In the case that a block does not compress, it is written on disk in uncompressed
form and represented as a full block. Thus the space occupied by a file is never
more than for the same file in a non-fragmented non-compressed file system.
Note that the compressed size is only known just before it is written to disk.
Since a program which writes a file does not expect an out-of-space (ENOSPC)
condition to occur after a successful write() (or successful store for mapped
files), it is necessary to guarantee that space will be available when logical
The AIX Version 4.1 implementation will affect the statfs structure and the format
of file system disk addresses. For the purpose of BSD compatibility, the statfs
structure is changed to include a new field for fragment size. The new f_fsize
field specifies file system fragment size in bytes and replaces the existing
f_nlsdirtype field. f_nlsdirtype was originally placed in the statfs structure under
AIX Version 3.2 for the purpose of future National Language Support
functionality, but is no longer required. Since this field has gone unused, it is
unlikely that it is referenced by any application programs: replacing f_nlsdirtype
with f_fsize should not create any binary or source compatibility problems.
Under the AIX Version 4.1 implementation, a new type definition is introduced for
representing JFS disk addresses. The frag_t type describes the area of disk
covered by a disk address, specifying both the starting location and length of the
area. While disk addresses are internally represented within JFS data structures
as frag_ts, the externalized data structures that record file system disk
addresses, including the disk inode and indirect block structures, remain
unchanged and continue to represent disk addresses as unsigned longs. The
details of frag_t are hidden for the purpose of source compatibility. Although
these source code details will be hidden, the format of file system disk
addresses will be documented in the < j f s / i n o . h > header file and the AIX
Version 4.1 user publications. This is done to accommodate the limited set of
customers that require low-level file system details.
Changes are required to the command line interface for the crfs, mkfs, and lsfs
commands. The crfs and mkfs commands will accept new arguments for
fragment size,the number of bytes per inode (nbpi), and compress to select data
compression. If compression is specified, the fragment size must be less than
4096 bytes. The lsfs command will be changed to display the fragment size, nbpi
and compress values for JFS file systems under the -q flag.
The fscnt() system call is extended to add a new value for the command
parameter, FS_MOVEBLKS.
The implementation will provide full AIX Version 3.2 disk image compatibility for
file systems and file system logs, allowing AIX Version 3.2 file systems to be
mounted and accessed in the usual manner without requiring customers to
perform any migration activities. In addition, the provided disk image
Two binary compatibility issues exist for the AIX Version 4.1 implementation:
quota support and programs that access JFS file systems through device I/O.
Under disk quota support, a user′s or group′s current disk space usage is
maintained in dqblk structures within quota files in units of 1024 bytes blocks. To
correctly maintain a user′s or group′s current usage, a carry bit is added to the
dqblk structure to account for partial current usage blocks. Since this structure
is an existing AIX Version 3.2 and contains no reserved space, the carry bit is
placed in the existing dqb_btime field. This scheme for recording partial current
usage should not effect the majority of current disk quota users, including those
using 512 bytes fragments. In fact, the impacts are likely be limited to programs
that access disk quota information for file systems with 512 byte fragment
directly through quota files rather than through the quotactl() system call. If any
of these programs exist, they will not be able to understand the format of the
dqb_btime field.
This scheme also does not work for users that take a quota file from a file
system with 512 byte fragments and move it to a file system with a larger
fragment size. This is particularly true if the quota file is moved to a AIX Version
3.2 file system on a AIX Version 3.2 system since this old version of the JFS does
not understand the format of the dqb_btime field. However, moving a quota file
from one file system to another generally makes little sense, particularly when
the file system fragment sizes are different.
File systems are somewhat unique in that they maintain on disk data structures
that are visible to programs through device I/O. In fact, some existing user
applications may accesses AIX Version 3.2 JFS file systems in this manner to
examine and use the on disk data structures. However the format of AIX Version
4.1 file systems differs from that of AIX Version 3.2 file systems. As a result,
programs that accesses AIX Version 4.1 file systems in this manner may not
work.
The implementation will change a number of the JFS specific data structures that
are shipped to customers, including the superblock, the incore inode, disk inode.
In general, these data structures do not represent programming interfaces; they
are externalized by tradition. This means that the changes to these structures
should have little or no impact to the application or kernel extension
environments. In fact, the impacts are probably limited to programs that access
file systems through device I/O and kernel extensions that violate subsystem
layering principles, like AFS.
The superblock structure is changed to include four new fields. The first two
fields describe basic system parameters and consists of s_fragsize and
s_iagsize. The third new field s_compress specifies whether data compression
is in effect. The fourth new field is s_version and specifies the superblock
version number. In addition, a new magic number, fsv4magic is introduced for
AIX Version 4.1 file systems.
A new field, i_movedfrag, is added to the incore inode and is used in recording a
file system object′s old committed resource.
With data compression, the number of processor cycles will also be increased.
Using software, the number of cycles for compression will be approximately 50
cycles per byte, and for decompression 10 cycles per byte (e.g. for a 601, a
compression rate of 1.2 megabytes per second, and a decompression rate of 6
megabytes a second).
The impacts to application programs of this enhancement are minimal since few
applications have intimate knowledge of the internal organization or
implementation of the file system. However, applications that directly access
logical volumes would have to be enhanced to enable access to logical volumes
beyond the 2 gigabyte limit. Applications will continue to work properly without
modification within the 2 gigabyte limitation.
The following AIX Version 3.2 Supplemental device support has been integrated
into the AIX Version 4.1 standalone CD_ROM and tape packages:
• Fiber Distribution Data Interface Diagnostics Supplement
• Block Multiplexer Channel Diagnostics Supplement
• 128 Port Asynchronous Communications Subsystem Diagnostics Supplement
In lieu of implementing FFST on AIX, the current software service aids are
enhanced to provide First Failure Data Capture. AIX currently provides facilities
for error logging, operator notification and network notification. It would need to
add:
• Symptom string support − As the term implies, a symptom string is meant
to describe a problem symptomatically. It is a sequence of keyword/value
pairs. The keywords are chosen from a predefined group, and the values
are problem specific. The keywords and values should be chosen in such a
way as to make it possible to identify duplicate problems. Ideally, there
should be one and only one symptom string for each valid problem.
The following changes are being made to support symptom strings:
1. Adding calls to allow either application or kernel code to generate a
symptom string as well as error log data.
2. Updating errdemon to put this data into the error log.
3. Updating errdemon to provide notification based on the presence of
symptom data.
4. Adding symptom string support to errpt.
• Selective dump capability − Selective dumping is not supported at present
by FFDC.
AIX Version 4.1 will save output from the boot process into a log file. This
includes the output from the period preceding and following display
configuration. The boot log is a fixed-size (circular) file whose size is
configurable. The default size to expected to be large enough to hold the output
from the two most recent boots.
The boot log is maintained using the alog command which is a new command in
AIX Version 4.1. The alog command is a general-purpose logging program that
reads standard input and writes the output to standard output and copies it to a
fixed-size file at the same time. The alog command works with log files that are
specified on the command line, or with logs that are defined in the alog
configuration objects in ODM. (The ″a″ in alog stands for AIX.)
The log file name and the in-memory buffer size are configurable in AIX Version
4.1. New flags have been added to the errdemon command for these functions.
In AIX Version 4.1, there are a couple of changes to the Error Notify Class:
• There is a new field added to the Class. This field (en_symptom) is reserved
for future use.
• The system′s objects are specified by the Error_ID instead of the
Error_Label.
In AIX Version 4.1, a CORE_DUMP error log entry will include the basename of
the program that core dumped. A sample is installed in
/usr/lpp/bos/samples/findcore to show how to locate the core file using the
information provided by the error log. This also serves as a general-purpose
error notification sample.
The errpt in AIX Version 4.1 will not be able to format the AIX Version 3.2 error
log; however, a conversion tool (/usr/lpp/bos.sysmgt/convert_errlog) is provided
to convert the AIX Version 3.2 error log to AIX Version 4.1 error log. If a
customer uses the migration install, the conversion will be done automatically.
errupdate is enhanced to disallow setting the alert flag to True if any of the
message identifiers in the template are not in the range permitted by SNA
Generic Alert Architecture. A new flag ′-p′ is provided to allow the customers to
override this restriction (since it is valid for customers to send alerts with
non-architected message identifiers).
errpt has several new flags and a couple of changes to existing flags:
-J Error_Label Retrieve error log entries for the specified Error_Label.
-K Error_Label Retrieve all the error log entries except those for the specified
Error_Label.
-y Filename Allow the alternate error record template file specified by the
Filename.
-z Filename Allow the alternate error logging message file specified by the
Filename.
-T Type_list In addition to the existing types (PERM, TEMP, PERF, PEND,
UNKN), there is a new type INFO.
-d Class_list A new class U (undetermined) can be used in addition to the
existing classes, H(hardware), S(software), and O(operator
message.
errdead has one new -i flag to write to an alternate error log file.
errdemon has a couple of new flags and a few changes to existing flag:
-B BufferSize To configure the error log internal buffer.
-i Filename To write to an alternate log file.
-l Lists the values for the name and size for the error log file and
the value for the buffer size.
errclear has several new flags and a few changes to existing flags:
-J Error_Label Delete the error log entries specified by the Error_Label.
-K Error_Label Delete all the error log entries except those specified by the
Error_Label parameter.
-T Type_list In addition to the existing types (PERM, TEMP, PERF, PEND,
and UNKN), there is a new type INFO.
-d Class_list A new class U (undetermined) can be used in addition to the
existing classes, H(hardware), S(software), and O(operator)
message.
trace Changes
The previous trace log will be preserved when restarting trace. The original file
is renamed (e.g. trcfile.old) and new data is written to a new file.
Trace templates are not installed in /etc/trcfmt until the trace option is installed.
The trace templates for options that are not in the bos.rte package are now
shipped with each individual option and will be installed when the option is
installed.
In AIX Version 4.1, the default primary dump device will be the paging space.
This saves disk space by no longer requiring you to keep a dedicated section of
your disk for dumps. You may, however, elect to use a dedicated dump logical
volume as before. If you use the migration facilities of base installation, your
dump device will remain unchanged. In either case, you can manually set your
dump device using SMIT or:
sysdumpdev -p.
If your dump goes to paging space, it must of course be copied off before your
system comes up and starts using the paging space again. This is done at boot
time with the savecore command, which copies the dump off your paging space
before paging activity is started.
The -D (capital D) parameter indicates that if the copy fails, you are given a
chance to copy the dump to removable media using a menu interface. The -d
(small d) flag tells the system to just continue if the copy fails. In this case,
you′ll lose the dump, but this may be necessary in some cases. The default is
-D, allow copy to external media.
The dump is copied to the file vmcore.n, where n is a sequence number. Your
first dump will be vmcore.1, the second vmcore.2, etc.
The most frequent reason for a copy to fail is that there′s not enough space in
the filesystem for the dump. The same space calculations you applied to your
dump device in AIX Version 3.2 applies to the free space in the target directory.
You may, however, elect to force some free space to be left in the file system
Presently the ECS command is used to access IBMLink via IBM Information
Network. ECS is included as part of the base AIX Version 3 Operating System
for use on the IBM RISC System/6000 only.
7.6 Security
Password Management:
• Password triviality checking
• Settable password expiration warning
• Password re-use
• Password lifetime
Login Controls
• Time of day/day of week restriction - user
• Time of day/day of week restriction - port
• Invalidate user after ′ N′ failed login attempts
• Disable port after ′ N′ failed login attempts
• Login retry delay
• Login completion timeout
There is no new function visible to the end user of the system. There are new
APIs to support a converged set of authentication interfaces, and there are new
capabilities available to the system administrator with respect to configuring how
and where a user authenticates (assuming DCE is installed).
For DCE, the behavior of the AUTH1 per-user attribute (located in the
/etc/security/user file) has been modified.
The existing SMIT functions under the Security and Users menu will be modified
to include the DCE and C2 security enhancements. The SMIT changes fall into
three basic categories:
1. Users − In the dialogs, new attributes were added for per-user password
and login controls and for DCE authentication and password registry
methods. Two new menu items were also added on the Users menu for
locking/unlocking user accounts and changing a user′s password.
2. Passwords − A new menu was added to provide a dialog for password
controls in addition to the existing function of changing a user′s password.
3. Logins − A new menu item on the Security and Users menu was added to
allow access to new dialogs for login controls for users or ports.
PDT must be enabled via the pdt_config command before it starts collecting
configuration, availability, and performance data on a daily basis. The data is
maintained in a historical record for approximately one month. On a daily basis,
PDT generates a diagnostic report that is mailed to the adm userid and written
to /var/perf/tmp/PDT_REPORT.
The tool will operate continuously; thus, there will be no explicit action the user
need take to run the tool. Also, continuous operation will allow customer
support to access tool output, potentially allowing for a post-analysis of the
customer′s problem without the need for further data collection.
The Performance PMR Data Collection Tool will also make use of certain
performance tools (such as tprof, filemon, netpmon) if they are installed on the
system. However, installation of these tools is not a prerequisite; if they are not
installed, then only the data available from the base AIX commands (for
example, iostat) will be collected.