NAME
shutdown — terminate all processing
SYNOPSIS
/sbin/shutdown
[-h|-r]
[-y]
[-o]
[grace]
DESCRIPTION
The
shutdown
command is part of the HP-UX system operation procedures.
Its primary function is to terminate all currently running processes
in an orderly and cautious manner.
shutdown
can be used to put the system in single-user mode
for administrative purposes such as backup
or file system consistency checks (see
fsck(1M)),
and to halt or reboot the system.
By default,
shutdown
is an interactive program.
Options and Arguments
shutdown
recognizes the following options and arguments.
- -h
Shut down the system and halt.
- -r
Shut down the system and reboot automatically.
- -y
Do not require any interactive responses from the user.
(Respond
yes
or
no
as appropriate to all questions,
such that the user does not interact with the shutdown process.)
- -o
When executed on the cluster
server
in a diskless cluster environment, shutdown
the
server
only and do not reboot
clients.
If this argument is not entered the default behavior
is to reboot all
clients
when the
server
is shutdown.
- grace
Either a decimal integer
that specifies the duration in seconds of a grace period
for users to log off before the system shuts down,
or the word
now.
The default is 60.
If
grace
is either 0 or
now,
shutdown
runs more quickly, giving users very little time to log out.
If neither
-r
(reboot) nor
-h
(halt) is specified,
standalone
and
server
systems are placed in single-user state. Either
-r
(reboot) or
-h
(halt) must be specified for a
client;
shutdown to single-user state
is not allowed for a
client.
See
dcnodes(1M),
init(1M).
Shutdown Procedure
shutdown
goes through the following steps:
The
PATH
environment variable is reset to
/usr/bin:/usr/sbin:/sbin.
The
IFS
environment variable is reset to space, tab, newline.
The user is checked for authorization to execute the
shutdown
command.
Only authorized users can execute the
shutdown
command.
See FILES for more information on the
/etc/shutdown.allow
authorization file.
The current working directory is changed to the root directory
(/).
All file systems' super blocks are updated; see
sync(1M).
This must be done before rebooting the system
to ensure file system integrity.
The real user ID is set to that of
the superuser.
A broadcast message is sent to all users currently logged in on the system
telling them to log out.
The administrator can specify a message at this time;
otherwise, a standard warning message is displayed.
The next step depends on whether a system is
standalone,
a
server,
or a
client.
If the system is
standalone,
/sbin/rc
is executed to shut down subsystems,
unmount file systems,
and perform other tasks to bring the system to run level 0.
If the system is a
server,
the optional
-o
argument is used to determine if all
clients
in the cluster should also be rebooted. The default behavior (command
line parameter
-o
is not entered) is to reboot all
clients
using
/sbin/reboot;
entering
-o
results in the
server
only being rebooted and the
clients
being left alone. Then
/sbin/rc
is executed to shut down subsystems,
unmount file systems,
and perform other tasks to bring the system to run level 0.
If the system is a
client,
/sbin/rc
is executed to bring the system down to run-level 2, and then
/sbin/reboot
is executed. Shutdown to the single-user state is not
an allowed option for
clients.
The system is rebooted or halted by executing
/sbin/reboot
if the
-h
or
-r
option was chosen.
If the system was not a cluster client and the system was being brought
down to single-user state,
a signal is sent to the
init
process to change states (see
init(1M)).
DIAGNOSTICS
- device busy
This is the most commonly encountered error diagnostic,
and happens when a particular file system could not be unmounted; see
mount(1M).
- user not allowed to shut down this system
User is not authorized to shut down the system.
User and system must both be included in the authorization file
/etc/shutdown.allow.
EXAMPLES
Immediately reboot the system and run HP-UX again:
Halt the system in 5 minutes (300 seconds)
with no interactive questions and answers:
Go to run-level
s
in 10 minutes:
FILES
- /etc/shutdown.allow
Authorization file.
The file contains lines that consist of a system host name
and the login name of a user who is authorized to reboot or halt the system.
A superuser's login name must be included in
this file in order to execute
shutdown.
However, if the file is missing or of zero length, the
root
user can run the
shutdown
program to bring the system down.
This file does not affect authorization
to bring the system down to single-user state for maintenance purposes;
that operation is permitted only when invoked by
a superuser.
A comment character,
#,
at the beginning of a line
causes the rest of the line to be ignored
(comments cannot span multiple lines without additional comment characters).
Blank lines are also ignored.
The wildcard character
+
can be used in place of a host name or a user name
to specify all hosts or all users, respectively (see
hosts.equiv(4)).
For example:
# user1 can shut down systemA and systemB
systemA user1
systemB user1
# root can shut down any system
+ root
# Any user can shut down systemC
systemC +
WARNINGS
The user name compared with the entry in the
shutdown.allow
file is obtained using
getlogin()
or, if that fails, using
getpwuid()
(see
getlogin(3C)
and
getpwent(3C)).
The hostname in
/etc/shutdown.allow
is compared with the hostname obtained using
gethostbyname()
(see
gethostent(3N)).
shutdown
must be executed from a directory on the root volume, such as the
/
directory.
The maximum broadcast message that can be sent
is approximately 970 characters.
When executing
shutdown
on an NFS diskless cluster server and the
-o
option is not entered, clients of the server will be rebooted.
No clients should be individually rebooted or shutdown while the
cluster is being shutdown.
SEE ALSO
dcnodes(1M),
fsck(1M),
init(1M),
killall(1M),
mount(1M),
reboot(1M),
sync(1M),
dcnodes(3X),
gethostent(3N),
getpwent(3C),
hosts.equiv(4).
