freebsd-nq/sbin/init/init.8
Andriy Gapon 5fda0d60c1 add ability to set watchdog timeout for a shutdown
This change allows to specify a watchdog(9) timeout for a system
shutdown.  The timeout is activated when the watchdogd daemon is
stopped.  The idea is to a prevent any indefinite hang during late
stages of the shutdown.  The feature is implemented in rc.d/watchdogd,
it builds upon watchdogd -x option.

Note that the shutdown timeout is not actiavted when the watchdogd
service is individually stopped by an operator.  It is also not
activated for the 'shutdown' to the single-user mode.  In those cases it
is assumed that the operator knows what they are doing and they have
means to recover the system should it hang.

Significant subchanges and implementation details:
- the argument to rc.shutdown, completely unused before, is assigned to
  rc_shutdown variable that can be inspected by rc scripts
- init(8) passes "single" or "reboot" as the argument, this is not
  changed
- the argument is not mandatory and if it is not set then rc_shutdown is
  set to "unspecified"
- however, the default jail management scripts and jail configuration
  examples have been updated to pass "jail" to rc.shutdown, just in case
- the new timeout can be set via watchdogd_shutdown_timeout rc option
- for consistency, the regular timeout can now be set via
  watchdogd_timeout rc option
- watchdogd_shutdown_timeout and watchdogd_timeout override timeout
  specifications in watchdogd_flags
- existing configurations, where the new rc options are not set, should
  keep working as before

I am not particularly wed to any of the implementation specifics.
I am open to changing or removing any of them as long as the provided
functionality is the same (or very close) to the proposed one.
For example, I think it can be implemented without using watchdogd -x,
by means of watchdog(1) alone.  In that case there would be a small
window between stopping watchdogd and running watchdog, but I think that
that is acceptable.

Reviewed by:	bcr (man page changes)
MFC after:	5 weeks
Relnotes:	yes
Differential Revision: https://reviews.freebsd.org/D21221
2019-10-03 11:23:10 +00:00

459 lines
12 KiB
Groff

.\" Copyright (c) 1980, 1991, 1993
.\" The Regents of the University of California. All rights reserved.
.\"
.\" This code is derived from software contributed to Berkeley by
.\" Donn Seeley at Berkeley Software Design, Inc.
.\"
.\" Redistribution and use in source and binary forms, with or without
.\" modification, are permitted provided that the following conditions
.\" are met:
.\" 1. Redistributions of source code must retain the above copyright
.\" notice, this list of conditions and the following disclaimer.
.\" 2. Redistributions in binary form must reproduce the above copyright
.\" notice, this list of conditions and the following disclaimer in the
.\" documentation and/or other materials provided with the distribution.
.\" 3. Neither the name of the University nor the names of its contributors
.\" may be used to endorse or promote products derived from this software
.\" without specific prior written permission.
.\"
.\" THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
.\" ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
.\" IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
.\" ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
.\" FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
.\" DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
.\" OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
.\" HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
.\" LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
.\" OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
.\" SUCH DAMAGE.
.\"
.\" @(#)init.8 8.3 (Berkeley) 4/18/94
.\" $FreeBSD$
.\"
.Dd August 6, 2019
.Dt INIT 8
.Os
.Sh NAME
.Nm init
.Nd process control initialization
.Sh SYNOPSIS
.Nm
.Nm
.Oo
.Cm 0 | 1 | 6 |
.Cm c | q
.Oc
.Sh DESCRIPTION
The
.Nm
utility
is the last stage of the boot process.
It normally runs the automatic reboot sequence as described in
.Xr rc 8 ,
and if this succeeds, begins multi-user operation.
If the reboot scripts fail,
.Nm
commences single-user operation by giving
the super-user a shell on the console.
The
.Nm
utility may be passed parameters
from the boot program to
prevent the system from going multi-user and to instead execute
a single-user shell without starting the normal daemons.
The system is then quiescent for maintenance work and may
later be made to go to multi-user by exiting the
single-user shell (with ^D).
This
causes
.Nm
to run the
.Pa /etc/rc
start up command file in fastboot mode (skipping disk checks).
.Pp
If the
.Em console
entry in the
.Xr ttys 5
file is marked
.Dq insecure ,
then
.Nm
will require that the super-user password be
entered before the system will start a single-user shell.
The password check is skipped if the
.Em console
is marked as
.Dq secure .
Note that the password check does not protect from variables
such as
.Va init_script
being set from the
.Xr loader 8
command line; see the
.Sx SECURITY
section of
.Xr loader 8 .
.Pp
If the system security level (see
.Xr security 7 )
is initially nonzero, then
.Nm
leaves it unchanged.
Otherwise,
.Nm
raises the level to 1 before going multi-user for the first time.
Since the level cannot be reduced, it will be at least 1 for
subsequent operation, even on return to single-user.
If a level higher than 1 is desired while running multi-user,
it can be set before going multi-user, e.g., by the startup script
.Xr rc 8 ,
using
.Xr sysctl 8
to set the
.Va kern.securelevel
variable to the required security level.
.Pp
If
.Nm
is run in a jail, the security level of the
.Dq host system
will not be affected.
Part of the information set up in the kernel to support a jail
is a per-jail security level.
This allows running a higher security level inside of a jail
than that of the host system.
See
.Xr jail 8
for more information about jails.
.Pp
In multi-user operation,
.Nm
maintains
processes for the terminal ports found in the file
.Xr ttys 5 .
The
.Nm
utility reads this file and executes the command found in the second field,
unless the first field refers to a device in
.Pa /dev
which is not configured.
The first field is supplied as the final argument to the command.
This command is usually
.Xr getty 8 ;
.Nm getty
opens and initializes the tty line
and
executes the
.Xr login 1
program.
The
.Nm login
program, when a valid user logs in,
executes a shell for that user.
When this shell
dies, either because the user logged out
or an abnormal termination occurred (a signal),
the cycle is restarted by
executing a new
.Nm getty
for the line.
.Pp
The
.Nm
utility can also be used to keep arbitrary daemons running,
automatically restarting them if they die.
In this case, the first field in the
.Xr ttys 5
file must not reference the path to a configured device node
and will be passed to the daemon
as the final argument on its command line.
This is similar to the facility offered in the
.At V
.Pa /etc/inittab .
.Pp
Line status (on, off, secure, getty, or window information)
may be changed in the
.Xr ttys 5
file without a reboot by sending the signal
.Dv SIGHUP
to
.Nm
with the command
.Dq Li "kill -HUP 1" .
On receipt of this signal,
.Nm
re-reads the
.Xr ttys 5
file.
When a line is turned off in
.Xr ttys 5 ,
.Nm
will send a SIGHUP signal to the controlling process
for the session associated with the line.
For any lines that were previously turned off in the
.Xr ttys 5
file and are now on,
.Nm
executes the command specified in the second field.
If the command or window field for a line is changed,
the change takes effect at the end of the current
login session (e.g., the next time
.Nm
starts a process on the line).
If a line is commented out or deleted from
.Xr ttys 5 ,
.Nm
will not do anything at all to that line.
.Pp
The
.Nm
utility will terminate multi-user operations and resume single-user mode
if sent a terminate
.Pq Dv TERM
signal, for example,
.Dq Li "kill \-TERM 1" .
If there are processes outstanding that are deadlocked (because of
hardware or software failure),
.Nm
will not wait for them all to die (which might take forever), but
will time out after 30 seconds and print a warning message.
.Pp
The
.Nm
utility will cease creating new processes
and allow the system to slowly die away, if it is sent a terminal stop
.Pq Dv TSTP
signal, i.e.\&
.Dq Li "kill \-TSTP 1" .
A later hangup will resume full
multi-user operations, or a terminate will start a single-user shell.
This hook is used by
.Xr reboot 8
and
.Xr halt 8 .
.Pp
The
.Nm
utility will terminate all possible processes (again, it will not wait
for deadlocked processes) and reboot the machine if sent the interrupt
.Pq Dv INT
signal, i.e.\&
.Dq Li "kill \-INT 1".
This is useful for shutting the machine down cleanly from inside the kernel
or from X when the machine appears to be hung.
.Pp
The
.Nm
utility will do the same, except it will halt the machine if sent
the user defined signal 1
.Pq Dv USR1 ,
or will halt and turn the power off (if hardware permits) if sent
the user defined signal 2
.Pq Dv USR2 .
.Pp
When shutting down the machine,
.Nm
will try to run the
.Pa /etc/rc.shutdown
script.
This script can be used to cleanly terminate specific programs such
as
.Nm innd
(the InterNetNews server).
If this script does not terminate within 120 seconds,
.Nm
will terminate it.
The timeout can be configured via the
.Xr sysctl 8
variable
.Va kern.init_shutdown_timeout .
.Pp
.Nm init
passes
.Dq Li single
as the argument to the shutdown script if return to single-user mode
is requested.
Otherwise,
.Dq Li reboot
argument is used.
.Pp
The role of
.Nm
is so critical that if it dies, the system will reboot itself
automatically.
If, at bootstrap time, the
.Nm
process cannot be located, the system will panic with the message
.Dq "panic: init died (signal %d, exit %d)" .
.Pp
If run as a user process as shown in the second synopsis line,
.Nm
will emulate
.At V
behavior, i.e., super-user can specify the desired
.Em run-level
on a command line, and
.Nm
will signal the original
(PID 1)
.Nm
as follows:
.Bl -column Run-level SIGTERM
.It Sy "Run-level Signal Action"
.It Cm 0 Ta Dv SIGUSR1 Ta "Halt"
.It Cm 0 Ta Dv SIGUSR2 Ta "Halt and turn the power off"
.It Cm 0 Ta Dv SIGWINCH Ta "Halt and turn the power off and then back on"
.It Cm 1 Ta Dv SIGTERM Ta "Go to single-user mode"
.It Cm 6 Ta Dv SIGINT Ta "Reboot the machine"
.It Cm c Ta Dv SIGTSTP Ta "Block further logins"
.It Cm q Ta Dv SIGHUP Ta Rescan the
.Xr ttys 5
file
.El
.Sh KERNEL ENVIRONMENT VARIABLES
The following
.Xr kenv 2
variables are available as
.Xr loader 8
tunables:
.Bl -tag -width indent
.It Va init_chroot
If set to a valid directory in the root file system, it causes
.Nm
to perform a
.Xr chroot 2
operation on that directory, making it the new root directory.
That happens before entering single-user mode or multi-user
mode (but after executing the
.Va init_script
if enabled).
This functionality has generally been eclipsed by rerooting.
See
.Xr reboot 8
.Fl r
for details.
.It Va init_exec
If set to a valid file name in the root file system,
instructs
.Nm
to directly execute that file as the very first action,
replacing
.Nm
as PID 1.
.It Va init_script
If set to a valid file name in the root file system,
instructs
.Nm
to run that script as the very first action,
before doing anything else.
Signal handling and exit code interpretation is similar to
running the
.Pa /etc/rc
script.
In particular, single-user operation is enforced
if the script terminates with a non-zero exit code,
or if a SIGTERM is delivered to the
.Nm
process (PID 1).
This functionality has generally been eclipsed by rerooting.
See
.Xr reboot 8
.Fl r
for details.
.It Va init_shell
Defines the shell binary to be used for executing the various shell scripts.
The default is
.Dq Li /bin/sh .
It is used for running the
.Va init_exec
or
.Va init_script
if set, as well as for the
.Pa /etc/rc
and
.Pa /etc/rc.shutdown
scripts.
The value of the corresponding
.Xr kenv 2
variable is evaluated every time
.Nm
calls a shell script, so it can be changed later on using the
.Xr kenv 1
utility.
In particular, if a non-default shell is used for running an
.Va init_script ,
it might be desirable to have that script reset the value of
.Va init_shell
back to the default, so that the
.Pa /etc/rc
script is executed with the standard shell
.Pa /bin/sh .
.Sh FILES
.Bl -tag -width /var/log/init.log -compact
.It Pa /dev/console
system console device
.It Pa /dev/tty*
terminal ports found in
.Xr ttys 5
.It Pa /etc/ttys
the terminal initialization information file
.It Pa /etc/rc
system startup commands
.It Pa /etc/rc.shutdown
system shutdown commands
.It Pa /var/log/init.log
log of
.Xr rc 8
output if the system console device is not available
.El
.Sh DIAGNOSTICS
.Bl -diag
.It "getty repeating too quickly on port %s, sleeping."
A process being started to service a line is exiting quickly
each time it is started.
This is often caused by a ringing or noisy terminal line.
.Bf -emphasis
Init will sleep for 30 seconds,
then continue trying to start the process.
.Ef
.It "some processes would not die; ps axl advised."
A process
is hung and could not be killed when the system was shutting down.
This condition is usually caused by a process
that is stuck in a device driver because of
a persistent device error condition.
.El
.Sh SEE ALSO
.Xr kill 1 ,
.Xr login 1 ,
.Xr sh 1 ,
.Xr ttys 5 ,
.Xr security 7 ,
.Xr getty 8 ,
.Xr halt 8 ,
.Xr jail 8 ,
.Xr rc 8 ,
.Xr reboot 8 ,
.Xr shutdown 8 ,
.Xr sysctl 8
.Sh HISTORY
An
.Nm
utility appeared in
.At v1 .
.Sh CAVEATS
Systems without
.Xr sysctl 8
behave as though they have security level \-1.
.Pp
Setting the security level above 1 too early in the boot sequence can
prevent
.Xr fsck 8
from repairing inconsistent file systems.
The
preferred location to set the security level is at the end of
.Pa /etc/rc
after all multi-user startup actions are complete.