Changed Commands in Section 1M

This section summarizes changes to commands in Section 1M of the HP-UX Reference that are not discussed elsewhere in this document.

Commands changed at 10.20:

fuser(1M)
mount(1M)
ncheck_hfs(1M)
unmount(1M)

Commands changed at 10.10:

bdf(1M)
df(1M)
frecover(1M)
savecore(1M)
shutdown(1M)

Commands changed at 10.01:

catman(1M)
savecore(1M)
swapon(1M)

Commands changed at 10.0:

config(1M)
dcopy(1M)
dodisk(1M)
fbackup(1M)
ff(1M)
fsck(1M)
fsdb(1M)
fuser(1M)
ioscan(1M)
labelit(1M)
lsdev(1M)
localedef(1M)
mkboot(1M)
mkfs(1M)
mount (1M), umount(1M)
ncheck(1M)
newfs(1M)
quot(1M)
quotacheck(1M)
sysdef(1M)
syslogd(1M)
volcopy(1M)
whodo(1M)

File System Administration Commands

The following commands now consist of a "wrapper", a script that determines the file system type and then executes the code appropriate to that type:

ff(1M)
fsck(1M)
fsdb(1M)
labelit(1M)
mkfs(1M)
mount(1M)
ncheck(1M)
newfs(1M)
quot(1M)
volcopy(1M)

See the entries for these commands later in this section. For details of the "wrapper", see the fs_wrapper(5) manpage on a 10.x system.

You will need to specify the file-system type, either on the command line for individual commands, or in /etc/fstab.

(If you do not, the commands will look in an HP-supplied file called /etc/default/fs; this may produce the result you want, but you should not rely on it.)

The file system-specific commands are stored in /sbin/fs/[FStype] or /usr/lbin/fs/[FStype]. They are not meant to be executed directly.

bdf(1M)

For 10.10:

bdf(1) will not report negative values for the field avail when the disk gets full. The value will always be greater than or equal to zero.
The %used field will not report values greater than 100. The value will always be less than or equal to 100.

For 10.01:

Supports multiple file systems
Output format changed
-t, -i options changed
-L no longer supported

Changed options:

   Option          Changed Behavior

   ==========================================================

   -t [type]       Extended to work with all supported file 

                   system types.

   -i              Displays -1 as number of free inodes for 

                   file systems that do not have a fixed 

                   number of inodes.

Option no longer supported:

   9.x Option      10.x Replacement    Comments

   ===========================================================



   -L              (none)              HP-UX "DUX" clusters  

                                       were not supported as  

                                       of 10.0.

Output format change:

The header now aligns with the statistics, and there is always a space between fields.

catman(1M)

catman has a new option at 10.01, -A <alt-root&>, to support client/server diskless environments. Under 10.01, diskless clients mount /usr read-only, and they cannot make changes to /usr/share/man. Therefore, catman(1M) can only be run on a server, and you may need to run catman with the -A option, depending on where the clients' manpages are located:

If the manpages are being served to a client, the client's manpages are located on the server in (for example):
/export/shared_roots/OS_700/usr/shared/man/
If the client and server are both 10.01, Series 700 machines, then the link /export/shared_roots/OS_700 -> / is present and the client and server have common manpages. In this case, simply run catman(1M) on the server.
If the client does not share manpages with the server, use the new -A <alt-root&> option to produce catman pages for a client. For example:
catman -A /export/shared_roots/OS_700
In this example, cat* manpages would be created in these (default manpaths):
/export/shared_roots/OS_700/usr/share/man/cat* /export/shared_roots/OS_700/usr/local/man/cat* /export/shared_roots/OS_700/usr/contrib/man/cat*
The whatis database would be created in /export/shared_roots/OS_700/usr/share/lib/whatis.

config(1M)

For 10.0:

Supports kernel build on Series 800 (replacing uxgen) as well as 700
Supports I/O convergence and new file system layout
Uses system file as input (similar to Series 700 dfile)
-a option no longer supported
New -s option
Only one swap device can be specified

config is the basis for all kernel builds on 10.x systems.




	NOTE: HP recommends you use `mk_kernel(1M)` or SAM (the menu-driven System Administration Manager) rather than `config` itself, to rebuild and install the kernel.

New option: -s

-s prevents config from calling make to build the kernel; conf.c and conf.mk are output.

Option no longer supported: -a

This option produced a script to help make device files. insf(1M) and mksf(1M) (supported on both the Series 700 and S800 platforms) are the best way to make device files on a 10.x system.

Input and Output Files

config accepts a system file as input, usually /stand/system. This file is similar in format to the 9.x Series 700 dfile, but the root and console device statements are no longer used. By default config writes its output files into the directory /stand/build (if the input file is /stand/system) and calls make(1) to compile and link a kernel, /stand/build/vmunix_test.

The /etc/master file was not supported as of 10.0. By default config now searches the directory /usr/conf/master.d and reads all the files in it for device specifications, tunable specifications, etc. See the master(4) manpage for a description of the master files.

dcopy(1M)




	NOTE: Release 10.10 will be the last release that `dcopy(1M)` will be supported.

For 10.0:

Complies with SVR4
Available on both Series 700 and 800

New options:

-F [FSType]: Specifies which file system type the command is to operate on. For 10.0, dcopy supports only HFS.
-V: Verbose. Echoes the completed command line.

As of 10.0, this command is stored in /usr/sbin/dcopy.

df(1M)

For 10.10:

Now conforms to XPG4 specifications.
Addition of the -P option. This option produces output that indicates the following:
file system name
Name of the system
total space
Size of the file system
space used
Total amount of space allocated to existing files in the file system
space free
Total amount of space available within the file system for creating new files by unprivileged users
percentage used
Percentage of the normally available space that is currently allocated to all files on the file system
file system root
Directory below which the file system hierarchy appears.
lf
New line character
The -P option is mutually exclusive with the -t option.
df(1M) does not report negative values for the field avail when the disk gets full. The value will always be greater than or equal to 0.
The %used field will not report values greater than 100 (previous releases did).

For 10.0:

Complies with SVR4
Supports multiple file systems

df now has both generic and file system-specific syntax and options. Many of the options were new for 10.0, including -F[FSType], -V (verbose), -e (files free), -g (statvfs report) -i (inodes report), -k (allocation in KBytes) -n (file system name), -o (file system-specific options), -v (blocks report) . For details, see the manpages or the HP-UX Reference, under df for generic usage and df_[FSType], for file system-specific usage--for example, look under df_hfs(1M) for HFS behavior.

Changed option:

                                                  New Option

           Old                  New               To Get Old

   Option  Behavior             Behavior          Behavior

   ============================================================

   -b      Report total number  Report number     -B

           of blocks for file   of KBytes free

           system swap, 

           allocated and free

You will need to specify the type of the file system df is to operate on--either on the command line or in /etc/fstab.

As of 10.0, this command is stored in /usr/bin/df. The 9.x version of the command is still available under /usr/old.

dodisk(1M)

This command has been modified to support multiple file systems. The changes will not affect existing (9.x) scripts.

fbackup(1M)

For 10.0:

Supports files that have extent attributes.
Supports loopback file systems (LOFS)

New options:

-E: Ignore extent attributes. (The default is to save them.)
-l: Back up LOFS files. (The default is to not to back them up.)

frecover(1M) will recover extent attributes. Other commands that support archiving extent attributes include cpio(1) and ``ftio(1). See the entries for those commands in this chapter, and the manpages.

ff(1M)

For 10.0:

Complies with SVR4.
Supports multiple file systems.
Available on both Series 700 and 800.

ff now has both generic and file system-specific syntax and options. For details, see the manpages or the HP-UX Reference, under ff for generic usage and ff_[FSType] for file system-specific usage--for example, look under ff_hfs(1M) for HFS behavior.

New options:

-F [FSType]: Specifies which file system type the command is to operate on.
-o [specific_options]: Specifies file system-specific options, if any.
-V: Verbose. Echoes the completed command line.

You will need to specify the type of the file system ff is to operate on--either on the commandline or in /etc/fstab. 10.x ff consists of a "wrapper" command which runs the version of the command appropriate to the file-system type specified. See the fs_wrapper(5) manpage for more information.

As of 10.0, this command is stored in /usr/sbin/ff.

frecover(1M)

For 10.10:

The frecover(1M) command might experience intermittent name cache failure. To offset this intermittent failure, the directory/file name is added to standard error even if this condition is encountered.

For 10.01:

Changed to support recovery of files that have extent attributes.

New option:

-E [extarg]: Specifies how to handle extent attributes and associated errors and warnings. (The default is to recover extent attributes.)

fsck(1M)

For 10.0:

Complies with SVID3
Supports multiple file systems

fsck now has both generic and file system-specific syntax and options. For details, see the manpages or the HP-UX Reference, under fsck for generic usage and fsck_[FSType] for file system-specific usage--for example, look under fsck_hfs(1M) for HFS behavior.

New options:

-F [FSType]: Specifies which file system type the command is to operate on.
-V: Verbose. Echoes the completed command line.
-m: Causes fsck to return zero if the file system is mountable, 32 if it needs additional checking, 33 if it is mounted, or a higher number if it is badly damaged.
-c 0: Eliminates use of the cache, reducing fsck's memory needs and allowing it to run on systems with 16Mb of memory. Not recommended for systems with more than 16Mb of memory because of the performance degradation.
-o [specific_options]: Specifies file system-specific options, if any.

Changed option:

                                                 New Option

           Old                  New              To Get Old

   Option  Behavior             Behavior         Behavior

   ============================================================



   -F      Check mounted        Specify file     -f

           file system          system type

You will need to specify the type of the file system fsck is to operate on--either on the command line or in /etc/fstab. 10.x fsck consists of a "wrapper" command which runs the version of the command appropriate to the file-system type specified. See the fs_wrapper(5) manpage for more information.

As of 10.0, this command is stored in /usr/sbin/fsck.

fsdb(1M)

For 10.0:

Complies with SVID3
Supports multiple file systems

fsdb now has both generic and file system-specific syntax and options. For details, see the HP-UX Reference, under fsdb for generic usage and fsdb_[FSType] for file system-specific usage--for example, look under fsdb_hfs(1M) for HFS behavior.

New options:

-F [FSType]: Specifies which file system type the command is to operate on.
-V: Verbose. Echoes the completed command line.
-o [specific_options]: Specifies file system-specific options, if any.

You will need to specify the type of the file system fsdb is to operate on--either on the commandline or in /etc/fstab. 10.x fsdb consists of a "wrapper" command which runs the version of the command appropriate to the file-system type specified. See the fs_wrapper(5) manpage for more information.

fuser(1M)

For 10.20:

fuser(1M) reports the processes that has memory-mapped the files that are passed as arguments to the process. The letter "m" will denote the memory-mapped processes.

fuser(1M) output format will be different if a file given as the argument to a process has been memory-mapped by any process. Programs or scripts that are dependent on the format will have to take into consideration the new letter "m" in the output.

For 10.0:

Complies with SVID3
Two new options
Report output changed

New options:

-c: Specifies that the directory is a mount point; any usage in that file system is reported.
-f: Causes target to be treated as a regular file, even if it is a file system.

Report output changes:

t: Indicates that this file is being used by a process as a text file.
o: Indicates open file.

These changes will affect scripts that explicitly direct stderr and stdout to the same file.

insf(1M), lssf(1M), mksf(1M), rmsf(1M)

For 10.0:

Reflect Series 700/800 converged IO; support HP-UX core drivers on 700
Look for device files in /dev by default
-l [lu] replaced by -I [instance]
New options -D, -q,-v

New options:

-D directory: Specifies the directory where the device files are to be found. Default is /dev.
-q: Quiet. Suppresses output.
-v: Verbose. Increases output.

Options no longer supported:

   9.x Option    10.x Replacement     Comments

   ============================================================



   -l [lu]       I [instance]         Logical Unit (LU) numbers

                                      no longer supported.

   -N [cnode]    (none)               HP-UX "DUX" clusters were  

                                      not supported as of 10.0.

Series 700 and 800 Convergence:

As of 10.0, these commands were supported in the same way on the Series 700 and 800, the only difference being which drivers are available. insf creates a device file for the "whole disk", not for each section or hard partition.

For more information, see “I/O Convergence” and the Configuring HP-UX for Peripherals manual.




	NOTE: `rmsf` will not remove most core drivers from the system, and will output warnings to that effect (unless you run it with `-q`).

ioscan(1M)

For 10.0:

Reflects Series 700/800 converged IO
-l [lu] replaced by -I [instance]
New option -F

As of 10.0, ioscan behaves the same way on the Series 700 and 800; the 9.x Series 800 command is the basis.

New options:

-F: Lists additional information which can be used to create customized listings via awk(1).

Changed option:

-f: Behaves the same on Series 700 and 800. Provides more information, in more readable form, than at 9.x.

Option no longer supported:

   9.x Option    10.x Replacement    Comments

   ===========================================================



   -l [lu]       I [instance]        Logical Unit (LU) numbers

                                     no longer supported.

For more information, see “I/O Convergence” and the Configuring HP-UX for Peripherals manual.

labelit(1M)

For 10.0:

Supports multiple file systems
New options -F, -V

New options:

-F [FSType]: Specifies which file-system type the command is to operate on. Required for tape.
-V: Verbose. Echoes the completed command line.

Except in the case of a tape, you can omit -F on the command line if you have defined the file-system type in /etc/fstab. 10.x labelit consists of a "wrapper" command which runs the version of the command appropriate to the file-system type specified. See the fs_wrapper(5) manpage for more information.

localedef(1M)

For 10.0:

localedef has undergone major changes. These are summarized below; see the manpage for details.

Grammar changes to comply with XPG4 and POSIX
Several new options (see below)
-d and -o options no longer supported
Locales are shared objects rather than binary files

New options:

-C [compiler_options]: Passes the specified options to the compiler that builds the locale.
-L [linker_options]: Passes the specified options to the linker that builds the locale.
-m [method_file]: Specifies a method file.
-v: Displays additional information for debugging.
-w: Displays duplicate-definition warnings.

Options no longer supported:

-d: No longer needed because locale description source files are provided on the system (under /usr/lib/nls/loc/src). See comments below on the effects of 10.x locales on 9.x code.
-o: Not needed because the -d option is no longer supported.




	NOTE: If you run `prepare(update_aid)` on a script that invokes `localedef`, you may get a message that the `-n` option is no longer supported. This is incorrect; `-n` is supported on 10.x. (See the Upgrading from HP-UX 9.x to 10.x manual for more information on `prepare`.)

Grammar changes:

Overall:
- Characters in portable codesets no longer need to be quoted (e.g., A need not be specified as 'A');
- The recommended approach is to use symbolic names for characters, as defined in the charmap files.
Header section: version, hpversion no longer supported.
LC_ALL category: context, direction moved to LC_CTYPE.
LC_CTYPE category: bytes_char, code_scheme, cswidth no longer supported.
LC_COLLATE category: modifier no longer supported.
LC_MONETARY category: crncystr no longer supported.

Effect of 10.x Locale Changes on 9.x Code

If 9.x Application Uses Customized 9.x Locales, and Linked Shared

If you have customized 9.x locales, you will need to modify the corresponding 10.x locales to get the same results on 10.x for applications that are linked shared. HP is providing a special 9.0-based localedef under /usr/old/usr/bin on 10.x systems for this purpose. This special localedef supports only the -d option to dump a locale.

Dump your 9.x-based locale, then apply the 9.x customizations to the 10.x locale source. This is safer than using the 9.x locale as the base and applying 10.x changes to it, and should be easier in most cases.

For example, to create a customized 10.x locale which in 9.x was based on the german locale, do the following:

Dump the customized 9.x locale:

On a 9.x system:

   localedef -d customized_locale_name > ldf1

On a 10.x system:

    /usr/old/usr/bin/localedef  -d customized_locale_name > ldf1

Dump the closest standard 9.x locale (german in this case):
- On a 9.x system:
  localedef -d german > ldf2
- On a 10.01 system:
  /usr/old/usr/bin/localedef -d german > ldf2
Identify the differences between the two dumps:
diff ldf1 ldf2
Use a text editor to apply the changes identified in the previous step to the 10.x german locale description file (/usr/lib/nls/loc/src/de_DE).

If 9.x Application Uses Standard 9.x Locales, and Linked Shared

Applications that were linked shared on 9.x will use the 10.x locales. End-users may see some differences because of changes to locales.

If 9.x Application Linked Archived

If the 9.x application was linked archived on 9.x, it will not be affected by 10.x locale changes, whether or not you customized 9.x locales; 9.x locales are still supported in 10.x, though this support will not necessarily continue beyond 10.01.

See the "Locales" section in the README for NLS in 10.01 for more information on changes affecting NLS locales.

lsdev(1M)

For 10.0:

Reflects Series 700/800 convergence
New options to report by driver name, class and major number

lsdev without options behaves the same as it did 9.x Series 800 systems.

The major difference between 9.x and 10.x on the Series 700 is that the device driver information is extracted from the running kernel instead of a static table. The Class column in the output is also new on the Series 700.

mkboot(1M)

For 10.0:

Reflects Series 700/800 convergence
All options supported on both Series 700 and Series 800
Supports three disk layouts on both Series 700 and Series 800: LVM, hard partitions, and whole-disk access.
Ensures that LVM target always has a label
Default boot program is the same for Series 700 and Series 800: /usr/lib/uxbootlf
By default uses the existing disk layout on the target, or LVM if no existing file system

New options:

-H: Specifies that target has hard partitions.
-W: Specifies that target has whole-disk layout.

Changed options:

-c: Additional checks: -i and -p options.
-f: Restricted to whole-disk layout; no change otherwise.
-h: Restricted to whole-disk layout; no change otherwise.

Options no longer supported:

-s: Series option; no longer needed.




	NOTE: LVM is the default layout if the target disk has no file system.

mkfs(1M)

For 10.0:

Complies with SVID3
Supports multiple file systems
Behaves the same on Series 700 and 800

mkfs now has both generic and file system-specific syntax and options. For details, see the HP-UX Reference, under mkfs(1M) for generic usage and mkfs_[FSType] for file system-specific usage--for example, look under mkfs_hfs for HFS behavior.

New options:

-F [FSType]: Specifies which file system type the command is to operate on.
-V: Verbose. Echoes the completed command line.
-m: Regenerates the command line used to create the file system.
-d: Forces mkfs to process a mounted system.
-o [specific_options]: Specifies file system specific-options, if any.

Changed option:

                                                 New Option 

           Old                  New              To Get Old 

   Option  Behavior             Behavior         Behavior

   ============================================================

   -F      Check mounted        Specify file     -d

           file system          system type

Changes from previous Series 700 or 800 behavior:

The following behavior is now common to both platforms:

Default rotational delay set to zero for all disk types.
Cylinder group summary area expanded to accommodate fsextend up to 4Gb.
Prototype file supported (to specify a file to be copied to the boot block). But mkboot(1M) is the normal way to do this.

You will need to specify the type of the file system mkfs is to operate on--either on the commandline or in /etc/fstab. 10.x mkfs consists of a "wrapper" command which runs the version of the command appropriate to the file-system type specified. See the fs_wrapper(5) manpage for more information.

As of 10.0, this command is stored in /usr/sbin/mkfs.

mount(1M), umount(1M)

For 10.20:

The mount(1M) and umount(1M) commands have been modified to:

support Multiple File System (MFS) in a wrapped paradigm
support large file systems (for HFS- and VxFS-specific mount commands only)
obsolete old options for all file system types

Summary of Change

Usage for mount:

           generic usage:



                mount [-l] [-p|-v]

                mount -a [-F FStype] [-eQ]

                mount [-F FStype] [-eQrV] [-o specific_options]

                      {special|directory}

                mount [-F FStype] [-eQrV] [-o specific_options]

                      special directory



           HFS-specific usage:



                mount [-l] [-p|-v]

                mount -a [-F hfs] [-eQ] [-f]

                mount [-F hfs] [-eQrV] [-f] [-o specific_options]

                      {special|directory}

                mount [-F hfs] [-eQrV] [-f] [-o specific_options]

                      special directory



           CDFS-specific usage:



                mount [-l] [-p|-v]

                mount -a [-F cdfs] [-eQ]

                mount [-F cdfs] [-eQrV] [-o specific_options]

                      {special|directory}

                mount [-F cdfs] [-eQrV] [-o specific_options]

                      special directory



           NFS-specific usage:



                mount [-l] [-p|-v]

                mount -a [-F nfs] [-eQ]

                mount [-F nfs] [-eQrV] [-o specific_options]

                      {host:path|directory}

                mount [-F nfs] [-eQrV] [-o specific_options]

                      host:path directory



           LOFS-specific usage:



                mount [-p|-v]

                mount -a [-F lofs] [-eQ]

                mount [-F lofs] [-eQrV] [-o specific_options]

                      {special_directory|directory}

                mount [-F lofs] [-eQrV] [-o specific_options]

                      special_directory directory



           VxFS-specific usage:



                mount [-l] [-v|-p]

                mount [-F vxfs] [-eQ] -a

                mount [-F vxfs] [-eQrV]

      [-o specific_options] {special | mount_point}

                mount [-F vxfs] [-eQrV]

      [-o specific_options] special mount_point

The following options are new for HFS- and VxFS-specific mount commands:

      -o largefiles        Attempts to enable the creation of files

                           greater than 2 GB in size.  File systems

                           have to be created or configured to enable

                           large files (see mkfs_hfs(1M) and 

                           fsadm_hfs(1M)).



      -o nolargefiles      Attempts to disable the creation of files

                           greater than 2 GB in size.  File systems

                           have to be created or configured to disable

                           large files (see mkfs_hfs(1M) and

                           fsadm_hfs(1M)).

The following options are obsolete: -u and -s.

-t type was obsoleted at 10.0. The replacement is -F FStype. The pre-10.0 interface-format is also obsolete. Instead, use the 10.0 interface-format.

Usage for umount:

           generic usage:



                umount [-v] [-V] {special|directory}

                umount -a [-F FStype] [-v]

  

           HFS-specific usage:



                umount -a [-F hfs] [-v]

                umount [-v] [-V] {special|directory}

           

           NFS-specific usage:



                umount -a [-F nfs] [-h host] [-v]

                umount [-v] [-V] {host:path|directory}

The following options are obsolete:

      -u                   The default mode enforces update (or -u's) 

                           functionality.



      -t type              The replacement is -F FStype.

The location of the mount and umount commands remains in /usr/sbin/mount and /usr/sbin/mount, respectively.

Impact

There are backward compatibility issues. The -u and -s options have been obsoleted. Also, the -t type option that was obsolete in 10.0 is no longer supported. It is replaced by -F FStype.

Performance

There is no significant impact to the performance of the mount and the umount commands. However, the performance of these commands is dependent on the type of the file system they operate on.

Limitations

You must have superuser access to mount/umount file systems using these commands.

For 10.0:

Comply with SVID3
Support multiple file systems
New options -F, -V, -v, -e, -o

mount and unmount now have both generic and file system-specific syntax and options. For details, see the manpages or HP-UX Reference, under mount and umount for generic usage, and mount_[FSType] and umount_[FSType] for file system-specific usage--for example, look under mount_hfs(1M) and umount_hfs(1M) for HFS behavior.

New options:

-F [FSType]

Specifies which file system type the command is to operate on.

-V

Verbose. Echoes the completed command line.

-e

Verbose. Writes a message to the standard output indicating which file system is being mounted. (Former -v behavior.)

-o [specific_options]

Specifies file-system-specific options, if any. New mount options for HFS:

no_fs_async: Use rigorous posting of file system metadata. (This is the default.)
fs_async: Use relaxed posting of file system metadata.
behind: Do the writes to disk asynchronously, where possible, without waiting for completion. Still the default on Series 700 systems.
delayed: Delay the writes to disk till the buffer needs to be reused. Still the default on Series 800 systems. delayed and behind are mutually exclusive.

Changed option (mount):

                                                 New Option

           Old                  New              To Get Old 

   Option  Behavior             Behavior         Behavior

   ============================================================

   -v      Indicate which       Same but also    -e

           file system is       indicate file

           being mounted or     system type

           unmounted

You will need to specify the type of the file system the command is to operate on--either on the command line or in /etc/fstab. 10.x mount and umount consist of a "wrapper" command which runs the version of the command appropriate to the file-system type specified. See the fs_wrapper(5) manpage for more information.

ncheck(1M)

For 10.20:

For 10.20, a new option is added to the HFS version of the ncheck command. The -S sector_range option reports on files that use the specified sector(s).

With the -S option, ncheck will print a table that reports the inode and pathname associated with the specified sectors. If no inode corresponds to the given sector, ncheck reports on what that sector is used for.

Performance

If the -S option is not used, the performance of ncheck is not impacted.

Compatibility

The output of ncheck has not changed for the case where -S is not specified. If the -S option is specified, ncheck prints a new table. Because -S is a new option, there are no backwards compatibility issues with respect to scripts that use ncheck. If the -S option is specified, the output will be different than the output of ncheck in previous releases.

For 10.0:

Complies with SVID3
Supports multiple file systems

ncheck now has both generic and file system-specific syntax and options. For details, see the manpages or the HP-UX Reference, under ncheck for generic usage and ncheck_[FSType] for file system-specific usage--for example, look under ncheck_hfs(1M) for HFS behavior.

New options:

-F [FSType]: Specifies which file system type the command is to operate on.
-V: Verbose. Echoes the completed command line.

Changed option:

-i [inodes]: List of inodes must be separated by commas; spaces are not allowed. This is directly contrary to 9.x and earlier usage, when the inode numbers were separated by blanks.

You will need to specify the type of the file system ncheck is to operate on--either on the command line or in /etc/fstab. 10.x ncheck consists of a "wrapper" command which runs the version of the command appropriate to the file-system type specified. See the fs_wrapper(5) manpage for more information.

As of 10.0, this command is stored in /usr/sbin/ncheck.

The 9.x version of the command is available on 10.x systems under /usr/old.

ncheck_hfs(1M)

For 10.20:

A new -S sector_range option has been added to the HFS version of the ncheck command. With the -S option, ncheck will print a table that reports the inode and pathname associated with the specified sectors. If no inode corresponds to the given sector, ncheck reports on what that sector is used for.

The output of ncheck has not changed if -S is not specified.

Because -S is a new option, there are no backwards compatibility issues with respect to scripts that use ncheck. If the -S option is specified, the output will be different than the output of ncheck in previous releases.

newfs(1M)

For 10.0:

Supports multiple file systems
/etc/disktab is no longer default source for disk parameters

newfs now has both generic and file system-specific syntax and options. For details, see the manpages or the HP-UX Reference, under newfs for generic usage and newfs_[FSType] for file system-specific usage--for example, look under newfs_hfs(1M) for HFS behavior.

New options:

-F [FSType]: Specifies which file system type the command is to operate on.
-V: Verbose. Echoes the completed command line.
-d: Tells newfs to disregard a mounted file system. (Former -F behavior.)
-R [swap]: Reserves [swap] megabytes of disk space for boot information or swap after the end of the file system.
-B: Reserves space for boot programs after the end of the file system.
-O [disk_type]: Tells newfs to look for the disk type in /etc/disktab. Parameters specified on the command line override corresponding values in /etc/disktab. Values not specified on the command line or in /etc/disktab are defaulted (see below).
-o [specific_options]: Specifies file system-specific options, if any.

Changed Options and Behavior

Changed options:

                                                 New Option

           Old                  New              To Get Old 

   Option  Behavior             Behavior         Behavior

   ============================================================

   -F      Disregard mounted    Specify file     -d

           file system          system type





   disk_   (Specified after    (See below)       -O

   type    device file

           for disk parameters)

In addition, the /etc/disktab entries supplied by HP no longer reserve swap and boot space at the end of the file system.

Use -O [disk_type] to force newfs to get disk geometry parameters from /etc/disktab.
Otherwise, newfs will use the defaults shown below.
Use -R and -B to reserve swap and boot space.

As of 10.0, /etc/disktab is the same file on the Series 700 and S800. In the case of some disk types, you will need to choose a Series 700- or Series 800-specific entry to get the same geometry as on 9.x.

newfs no longer requires you to specify disk_type on the command line, and, by default, does not use /etc/disktab to get the disk geometry parameters. Instead it uses these defaults:

   File system size:    disk size returned by DIOC_CAPACITY 

                        (minus swap/boot)

   Block size:          8192

   Fragment size:       1024

   Tracks per cylinder: calculated as shown below

   Sectors per track:   calculated as shown below

   RPM:                 3600

Tracks per cylinder and sectors per track are calculated according to the size of the file system, as follows:

   File System Size    Tracks/Cylinder        Sectors/Track

   ========================================================

   <= 500 MB            7                      22

   501 MB - 1 GB        12                     28

   > 1 GB               16                     39

The default value for bytes per inode has increased from 2048 to 6144.

You will need to specify the type of the file system newfs is to operate on--either on the command line or in /etc/fstab. 10.x newfs consists of a "wrapper" command which runs the version of the command appropriate to the file-system type specified. See the fs_wrapper(5) manpage for more information.

As of 10.0, this command is stored in /usr/sbin/newfs.

The 9.x version of newfs is available on 10.x systems as /usr/old/etc/newfs.

quot(1M)

For 10.0:

Supports multiple file systems
New options -V, -F

-F [FSType]: Specifies which file system type the command is to operate on.
-V: Prints out the usage string.

You will need to specify the type of the file system quot is to operate on--either on the command line or in /etc/fstab. 10.x quot consists of a "wrapper" command which runs the version of the command appropriate to the file-system type specified. See the fs_wrapper(5) manpage for more information.

quotacheck(1M)

For 10.0:

Supports multiple file systems
New options -V, -F

-F [FSType]: Specifies which file-system type the command is to operate on.
-V: Prints out the usage string.

savecore(1M)

For 10.10:

savecore now saves system crash dumps in a new format, which has many files for each crash dump, all saved in a subdirectory. Support for subset core files containing only user or kernel page has been removed.

Additionally, the following changes have been made to savecore(1M):

Each crash dump now has its own directory, a subdirectory of /var/adm/crash (or another directory specified on the command line).
The crash dump directory contains an INDEX file describing the dump, a multitude of small files (instead of one big one) containing the physical memory image, and file(s) containing the kernel. Any of these files except INDEX can be compressed.
savecore no longer offers the ability to save subsets of the core dump (for example, only user and kernel pages). All parts of the dump are saved. The relevant command line options (-i, -u, -k, -S, and -F) is ignored.
savecore always saves as much of the dump as it can given space limitations. (The previous version of savecore did not save anything unless the whole dump would fit.) The option -p is ignored.
The optional "system" command line parameter is ignored because it is no longer needed.
The options -z, -Z, -s, -r, and -m are added to control the parameters of saving a dump in the new format.
The options -D and -O are added to allow you to dictate where savecore should look for the dump in special cases such as early boot dumps. The ability of savecore to find the dump without help in normal cases is signficantly enhanced.
Option -l is added to allow savecore to update system log files without physically saving the dump.

For more information on these changes, refer to the savecore(1M) and savecore(4) manpages.

Impact

Any and all programs and utilities that rely on the format of a core dump will need to change to reflect the new dump format being written by savecore. There are no supported non-HP core dump tools, so this should not have an impact. All HP-supported tools (such as adb) are changed to support the new format.

The removal of support for subset core dumps may imply a slight increase in the disk space used by core dumps for those who used this functionality.

Performance

If you have sufficient disk space that savecore does not perceive a need for compression, or if you explicitly disable compression, the performance of the new version of savecore will be unchanged from the old. If compression is enabled, performance will degrade significantly in favor of reduced disk usage.

Size Requirement

A crash dump for any given system will use incrementally more disk space (for the INDEX file), although space will be more than compensated by the space savings of compression. When compression is in use, virtual memory use of savecore will significantly increase.

For 10.0:

savecore has one new feature affecting both the Series 700 and 800, and a group of new features on the 700 that are already supported on Series 800 9.x systems.

New on Both Series 700 and 800

The corefile is no longer saved in the directory /tmp/syscore; the default directory as of 10.0 is /var/adm/crash. You can change this and other defaults by modifying variables in /etc/rc.config.d/savecore:

   Variable     Default                          Other 

   Name         Value (Meaning)     Description  Valid Values

   ===========================================================



   SAVECORE     1 (always           Save core    0 (do not 

                save core file)     or not?      save)

                                 



   SAVECORE_DIR /var/adm/crash      Directory    (any direct-

                                    where core-  ory; free 

                                    file will    space in file 

                                    be saved     system should

                                                 at least 

                                                 equal RAM)



   SAVE_PAGES   (blank) (save       Memory       i, u, k

                complete image)     pages to     (see "Smaller

                                    save         Corefile"

                                                 below)

New on Series 700

Features formerly supported only on the Series 800, but now available on the Series 700 as well, allow you to:

Reduce system bootup time.
Reduce the size of the corefile after a core dump.

Faster system bootup

10.x savecore supports multiple dump devices on both the Series 800 and 700. If the primary swap device (PSD) is not configured as a dump device, savecore will start running in the background immediately; otherwise it must save the dump image in the PSD before switching to background mode.

Smaller corefile

You can make savecore retrieve pages from the dump devices selectively by using the -u, -k or -i options. -u saves user and kernel pages; -k saves kernel pages only; -i saves the entire image if there's space. See the 10.x or (9.x Series 800) manpage for details. You can change the system defaults for these values by modifying /etc/rc.config.d/savecore as shown in the table above.

shutdown(1M)

For 10.10:

When executed on an NFSD cluster server, shutdown(1) can
- shut down all the clients in the NFSD cluster
- shut down only the cluster server and leave the clients untouched by using the -o option.
When executed on the server, warning messages can be sent to users on the clients about the impending shutdown.
Shutdown executed on clients in an NFSD cluster can be used to only reboot or halt the client. NFSD clients cannot be shut down to the single-user state.
The option -r or -h must be entered when executing a shutdown on an NFSD client.

For 10.0:

Follows new "V.4" paradigm
/etc/shutdown.d directory no longer supported
No cluster-wide shutdown: In an NFS Diskless cluster, executing shutdown(1M) on the cluster server does not shut down all cluster nodes.

The new shutdown and startup model replaces both the shutdown actions that were previously hardcoded into shutdown(1M) and the execution of scripts in directory /etc/shutdown.d.

For information about the new model, see the HP-UX 10.0 File System Layout White Paper and the HP-UX System Administration Tasks manual.

Changing Existing Scripts

If you have scripts in /etc/shutdown.d on a 9.x system, and you want to move them to a 10.x system, you will need to migrate them to the new method. For instructions, see the section on startup and shutdown scripts in the Upgrading from HP-UX 9.x to 10.x manual.

swapon(1M)

As of 10.01, swapon accepts arguments to the -m, -r, and -l flags that have units specified. The corresponding /etc/fstab entries are similarly changed.

Arguments to the -r, -m, and -l flags, if specified without units, are assumed to be in units of file system blocks. This maintains backward compatibility. However, these arguments can be suffixed with a k or M, which change the units to kilobytes or megabytes, respectively. The res=, min=, and lim= entries in /etc/fstab behave in the same manner.




	NOTE: Although one of the command line forms for `swapon` may be marked as obsolete in the manpage, it has not been removed. It may be removed in a future release.

sysdef(1M)

For 10.0:

Displays ranges and units of tunable parameters
Flags parameters which can change while system is running
Shows boot-time and run-time values if they differ
No longer displays swap or memory information
Output format changed

Use swapinfo(1M) to get the swap information that used to be supplied by sysdef. Use vmstat(1) to get memory information.

syslogd(1M)

For 10.0:

Kernel messages logged via /dev/klog
New -D option

New option:

-D: Prevent the kernel from printing messages directly onto the system console (route them through syslogd instead).

Kernel messages will be routed by default to /var/adm/syslog/syslog.log (which is the equivalent of 9.x /usr/adm/syslog), or to whatever other file you have specified for system logging. If you do not want kernel messages to appear in the logfile, specify a priority of kern.none.

unmount(1M)

For 10.20:

See “mount(1M), umount(1M)”.

volcopy(1M)

For 10.0:

Supported on the Series 700 as well as 800
Supports multiple file systems
New options -F, -V

New options:

-F [FSType]: Specifies which file-system type the command is to operate on. Required for tape.
-V: Verbose. Echoes the completed command line.

Except in the case of a tape, you can omit -F on the command line if you have defined the file-system type in /etc/fstab. 10.x volcopy consists of a "wrapper" command which runs the version of the command appropriate to the file-system type specified. See the fs_wrapper(5) manpage for more information.

whodo(1M)

For 10.0:

Complies with SVID3
Two new options -h and -l
Can report on a particular user

New options:

-h: Omit report header.
-l: Report in long form.

If invoked without options, whodo produces the same result as on 9.x.

Changed Commands in Section 1M

Technical documentation

» Table of Contents

» Index

File System Administration Commands

bdf(1M)

catman(1M)

config(1M)

Input and Output Files

dcopy(1M)

df(1M)

dodisk(1M)

fbackup(1M)

ff(1M)

frecover(1M)

fsck(1M)

fsdb(1M)

fuser(1M)

insf(1M), lssf(1M), mksf(1M), rmsf(1M)

ioscan(1M)

labelit(1M)

localedef(1M)

Effect of 10.x Locale Changes on 9.x Code

If 9.x Application Uses Customized 9.x Locales, and Linked Shared

If 9.x Application Uses Standard 9.x Locales, and Linked Shared

If 9.x Application Linked Archived

lsdev(1M)

mkboot(1M)

mkfs(1M)

mount(1M), umount(1M)

Summary of Change

Impact

Performance

Limitations

ncheck(1M)

Performance

Compatibility

ncheck_hfs(1M)

newfs(1M)

Changed Options and Behavior

quot(1M)

quotacheck(1M)

savecore(1M)

Impact

Performance

Size Requirement

New on Both Series 700 and 800

New on Series 700

Faster system bootup

Smaller corefile

shutdown(1M)

Changing Existing Scripts

swapon(1M)

sysdef(1M)

syslogd(1M)

unmount(1M)

volcopy(1M)

whodo(1M)