Use .Va for errno.

[dragonfly/vkernel-mp.git] / usr.sbin / dntpd / dntpd.8

.\" $DragonFly: src/usr.sbin/dntpd/dntpd.8,v 1.13 2007/06/26 02:40:20 dillon Exp $
.\"
.\" Copyright (c) 2005 The DragonFly Project.  All rights reserved.
.\"
.\" This code is derived from software contributed to The DragonFly Project
.\" by Matthew Dillon <dillon@backplane.com>
.\"
.\" 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 DragonFly Project 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 COPYRIGHT HOLDERS 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
.\" COPYRIGHT HOLDERS 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.
.\"
.Dd April 26, 2005
.Dt DNTPD 8
.Os
.Sh NAME
.Nm dntpd
.Nd Network time protocol client daemon
.Sh SYNOPSIS
.Nm
.Bk -words
.Op Fl dnqstFSQ
.Op Fl f Ar config_file
.Op Fl i Ar insane_deviation
.Op Fl l Ar log_level
.Op Fl T Ar nominal_poll
.Op Fl L Ar maximum_poll
.Op targets
.Ek
.Sh DESCRIPTION
The
.Nm
daemon will synchronize the system clock to one or more external NTP time
sources.  By default an initial coarse offset correction will be made if
time is off by greater than 2 minutes.  Additional sliding offset
corrections will be made if necessary.  Once sufficient information is
obtained,
.Nm
will also correct the clock frequency.  Over the long haul the frequency can
usually be corrected to within 2 ppm of the time source.  Offset errors can
typically be corrected to within 20 milliseconds, or within 1 millisecond of
a low latency time source.
.Pp
By default
.Nm
will load its configuration from
.Pa /etc/dntpd.conf
and run as a daemon (background itself).  If you re-execute
the binary it will automatically kill the currently running
.Nm
daemon.  If you run
.Nm
with the -Q option any currently running daemon will be killed and
no new daemon will be started.
.Pp
The following command line options are available:
.Bl -tag -width Fl
.It Fl d
Run in debug mode.  Implies
.Fl F ,
.Fl l Ar 99 ,
and
.Fl f Ar /dev/null
and logs to stderr instead of syslog.  The normal client code is run and
time corrections will be made.
.It Fl n
No-update mode.  No actual update is made any time the client would
otherwise normally update the system frequency or offset.
.It Fl q
Quiet mode.  Implies a logging level of 0.
.It Fl s
Issue a coarse offset correction on startup.  Normally a coarse offset
correction is only made when the time differential is greater than 2
minutes.  This option will cause the initial offset correction to be
a coarse correction regardless.  Note that the system will still not make
a correction unless the offset error is greater than 4 times the standard
deviation of the queries.
.It Fl t
Test mode.  Implies
.Fl F ,
.Fl l Ar 99 ,
.Fl n ,
and
.Fl f Ar /dev/null
and logs to stderr instead of syslog.  A single linear regression is
accumulated at the nominal polling rate and reported until terminated.
No time corrections are made.  This option is meant for testing only.
Note that frequency corrections based on internet time sources typically
require a long (10-30min) polling rate to be well correlated.
.It Fl F
Run in the foreground.  Unlike debug mode, this option will still log
to syslog.
.It Fl S
Do not set the time immediately on startup (default).
.It Fl Q
Terminate any running background daemon and exit.
.It Fl f Ar config_file
Specify the configuration file.  The default is
.Pa /etc/dntpd.conf .
.It Fl i Ar insane_deviation
Specify how much deviation is allowed in calculated offsets, in seconds.
Fractions may be specified.
A quorum of servers must agree with the one we select as being the best time
source in order for us to select that source.
The default deviation allowed is a fairly expansive 0.5 seconds.
Note that offset errors due to internet packet latency can
exceed 25ms (0.025).
.It Fl l Ar log_level
Specify the log level.  The default is 1.  All serious errors are logged
at log level 0.  Major time corrections are logged at log level 1.  All
time corrections and state changes are logged at log level 2.  Log level's
3 and 4 increase the amount of debugging information logged.
.It Fl T Ar nominal_poll
Set the nominal polling interval, in seconds.  This is the interval used
while the client is in acquisition mode.
The default is 300 seconds (5 minutes).
.It Fl L Ar maximum_poll
Set the maximum polling interval, in seconds.  This is the interval used
while the client is in maintenance mode, after it believes it has
stabilized the system's clock.
The default is 1800 seconds (30 minutes).
.It targets
Specify targets in addition to the ones listed in the config file.  Note
that certain options (-d, -t) disable the config file, and you can specify
a configuration file of
.Pa /dev/null
if you want to disable it otherwise.
.El
.Sh IMPLEMENTATION NOTES
.Nm
runs two linear regressions for each target against the uncorrected system
time.  The two linear regressions are staggered so the second one is stable
and can replace the first one once the first's sampling limit has been
reached.
The second linear regression is also capable of overriding the first if
the target changes sufficiently to invalidate the first's correlation.
.Pp
The linear regression is a line-fitting algorithm which allows us to
calculate a running Y-intercept, slope, and correlation factor.  The
Y-intercept is currently not used but can be an indication of a shift in
the time source.  The slope basically gives us the drift rate which in
turn allows us to correct the frequency.  The correlation gives us a
quality indication, with 0 being the worst and \(+- 1.0 being the best.
.Pp
A standard deviation is calculated for offset corrections.  A standard
deviation gives us measure of the deviation from the mean of a set of
samples.
.Nm
uses the sum(offset_error) and sum(offset_error^2) method to calculate
a running standard deviation.   The offset error relative to the
frequency-corrected real time is calculated for each sample.  Note that
this differs from the uncorrected offset error that the linear regression
uses to calculate the frequency correction.
.Pp
In order to make a frequency correction a minimum of 8 samples and a
correlation \(>= 0.99, or 16 samples and a correlation \(>= 0.96 is required.
Once these requirements are met a frequency correction will typically be
made each sampling period.  Frequency corrections do not 'jump' the system
time or otherwise cause fine-time computations to be inaccurate and thus
can pretty much be made at will.
.Pp
In order to make an offset correction a minimum of 4 samples is required
and the standard deviation must be less than \(14 the current calculated
offset error.  The system typically applies offset corrections slowly over
time.  The algorithm will make an offset correction whenever these standards
are met but the fact that the offset error must be greater than 4 times the
standard deviation generally results in very few offset corrections being
made once time has been frequency-corrected.
.Nm
will not attempt to make a followup offset correction until the system
has completed applying the previous offset correction, as doing so would
cause a serious overshoot or undershoot.  It is possible to use a more
sophisticated algorithm to take running offset corrections into account
but we do not do that (yet).
.Pp
.Nm
maintains an operations mode for each target.  An initial 6 samples are taken
at 5 second intervals, after which samples are taken at 5 minute intervals.
If the time source is deemed to be good enough (using fairly relaxed
correlation and standard deviation comparisons) the polling interval is
increased to 30 minutes.  Note that long intervals are required to get good
correlations from internet time sources.
.Pp
If a target stops responding to NTP requests the operations mode goes into a
failed state which polls the target at the nominal polling rate
(e.g. 5 minutes).  Once re-acquired
.Nm
will either go back to the 5-second startup mode or to the 5-minute
acquisition mode depending on how long the target was in the failed state.
.Sh TIME SYNCHRONIZATION ISSUES
If the system clock is naturally off-frequency
.Nm
will be forced to make several offset corrections before it gets enough data
to make a frequency correction.  Once the frequency has been corrected
.Nm
can typically keep the time synchronized to within 1-20 milliseconds depending
on the source and both the number of offset corrections and the size of the
offset corrections should be significantly reduced.
.Pp
It will take up to 30 seconds for
.Nm
to make the initial coarse offset correction.  It can take anywhere from
5 minutes to 3 hours for
.Nm
to make the initial frequency correction, depending on the time source.
Internet time sources require long delays between samples to get a high
quality correlation in order to issue a frequency correction.
.Pp
It is difficult to calculate the packet latency for an internet time source
and in some cases this can result in time sources which disagree as much as
20ms with each other.  If you specify multiple targets and run in
debug or a high-logging mode you may observe this issue.
.Sh MULTIPLE SERVERS AND DNS ROUND ROBINS
Multiple servers may be specified in the configuration file.  Pool domains
are supported and the same domain name may be specified several times to
connect to several different targets within the pool.  Your DNS server
must rotate IPs for this to work properly (all
.Ux
name servers will rotate IPs).
.Nm
will automatically weed out any duplicate IPs.
.Pp
When two or more time sources are configured,
.Nm
will do a quorum-based sanity check on its best pick and fail the server if
its offset deviates significantly from other servers.
.Pp
If a server fails,
.Nm
will relookup its domain name and attempt to reconnect to it.
To avoid overloading servers due to packet routing snafus, reconnections
can take upwards of an hour to cycle.
.Sh CONFIGURATION FILE
The
.Pa /etc/dntpd.conf
file contains a list of servers in the 'server <servername>' format, one
per line.  Any information after a '#' is assumed to be a comment.  Any
number of servers may be specified but it is usually wasteful to have more
than four.
.Sh FILES
.Bl -tag -compact
.It Pa /var/run/dntpd.pid
When started as a daemon,
.Nm
stores its pid in this file.  When terminating a running
.Nm
this file is used to obtain the pid.
.Pp
.It Pa /etc/dntpd.conf
The default configuration file.
.El
.Sh AUTHORS
This program was written by Matthew Dillon.
.Sh BUGS
An algorithm is needed to deal with time sources with packet-latency-based
offset errors.
.Pp
The offset correction needs to be able to operate while a prior offset
correction is still in-progress.
.Pp
We need to record the frequency correction in a file which is then read on
startup, to avoid having to recorrect the frequency from scratch every
time the system is rebooted.