source: trunk/athena/bin/desync/desync.1 @ 25710

.\" $Id: desync.8,v 1.3 2000-09-30 21:08:29 rbasch Exp $
.\"
.\" Copyright 1995, 1996, 1997 by the Massachusetts Institute of Technology.
.\"
.\" Permission to use, copy, modify, and distribute this
.\" software and its documentation for any purpose and without
.\" fee is hereby granted, provided that the above copyright
.\" notice appear in all copies and that both that copyright
.\" notice and this permission notice appear in supporting
.\" documentation, and that the name of M.I.T. not be used in
.\" advertising or publicity pertaining to distribution of the
.\" software without specific, written prior permission.
.\" M.I.T. makes no representations about the suitability of
.\" this software for any purpose.  It is provided "as is"
.\" without express or implied warranty.
.\"
.TH DESYNC 1 "3 August 2012"
.SH NAME
desync \- desynchronize timed jobs on networks
.SH SYNOPSIS
.B desync
[
.B \-t
timefile ] [ range ]

.B desync -c
hours [ range [ other arguments ] ]
.SH DESCRIPTION
.I desync
is a tool which sleeps a random (hostname seeded) period of time (up
to an hour, by default) in order to skew the network load from several
machines running
.I cron
jobs that would otherwise be synchronized.

An invocation would appear something like

.nf
     42 * * * *          desync; /usr/lib/sendmail -q
.fi

in a crontab file.
.SH OPTIONS
.TP 8
.B range
This optional range specifies, in seconds, the maximum value for the
randomly selected sleep time.  The default value is 3600 (except in
"crontab" mode, see below).
.TP 8
.B \-h name
Use
.I name
to seed the random number generator, instead of using the current host
name.  This option is generally only useful in conjunction with the
.B \-n
option.
.TP 8
.B \-n
Instead of sleeping, this option causes
.I desync
to write the number of seconds it would sleep to standard output, or,
if the
.B \-t
option is also specified, the number of seconds remaining until the
timer in
.I timefile
expires.
.TP 8
.B \-t timefile
This option changes the behavior of
.I desync
to make it useful for desynchronizing jobs when sleeping is not
appropriate.  If
.I timefile
does not exist,
.I desync
selects a random sleep time using the normal method, adds it to the
current time value, writes the result into
.IR timefile ,
and exits with status 1.  If
.I timefile
does exist and the current time is less than the time value listed in
.IR timefile ,
then
.I desync
takes no action and exits with status 1.  If
.I timefile
exists and the current time is equal to or greater than the time value
listed in
.IR timefile ,
then
.I desync
unlinks
.I timefile
and exits with status 0.  This feature can be used in shell code of
the following form:

.nf
        if desync -t /tmp/mytimefile 3600; then
                echo "Starting the job."
        else
                echo "Putting it off until later."
        fi
.fi

.TP 8
.B \-c hours
This option changes the behavior of
.I desync
and causes it to generate output suitable for use in a crontab file.  It
also changes the units of 
.I range
to minutes (from seconds), changes the default value to 60 and enforces
a maximum value of 1439.  This is useful on modern Linux distributions,
where sleeping for extended periods inside a cron job can confuse power
management software or packages such as ConsoleKit, and running a cron
job every 5 minutes to see if desync thinks it is "time to run" is
undesirable.  Since desync will generate the same value each time on the
same machine, it is fine to generate crontab files in a package's
post-install script, for example.  In this mode, desync will output the
crontab fields and then any additional arguments you supply.  So to
generate a crontab with a randomized job, one might do something like
this:

.nf
        for i in 2 4 8 14 20; do
            desync -c $i 120 root /etc/athena/update_ws >> /etc/cron.d/update
        done
.fi


.SH SEE ALSO
cron(8)
.SH AUTHOR
Craig Fields, MIT Information Systems
.br
Greg Hudson, MIT Information Systems
.br
Debathena Project
.br
Copyright (c) 1995, 1996, 1997, 2012 Massachusetts Institute of Technology