#!/bin/ksh
#
# mirror
# 
# AUTHOR:
#   Dan Harkless                                                    <opensource@
#   http://harkless.org/dan/software/                              harkless.org>
# 
# COPYRIGHT:
#   This file is Copyright (C) 1999 by Dan Harkless, and is released under the
#   GNU General Public License <http://www.gnu.org/copyleft/gpl.html>.
#
# USAGE:
#   % mirror <directory> [<user1>@]<host1> [[<user2>@]<host2>...]
# 
# DESCRIPTION:
#   The only single command UNIX provides for mirroring a directory onto another
#   machine is the rdist command, kind of the redheaded step-child of the
#   r<command> family.
#
#   It's not a bad command, but there's a bit of a learning curve for its
#   "distfiles".  And the main problem with it from my standpoint is that it
#   crawls through the directory structure and replaces files one-by-one.  This
#   is unacceptable for directory trees like websites where files refer to one
#   another.  If file A links to a brand new file B, and you transfer file A
#   over before file B, you'll have a window during which the link in file A
#   will appear to be broken.
#
#   A much preferable way to mirror "live" directories like this is to leave the
#   remote machine's old hierarchy alone, while building a new, updated
#   hierarchy off to the side there.  Once the new hierarchy has been completely
#   built (or simply transferred over), the old top directory is switched out
#   and the new one is put in its place.  At this point the old directory
#   hierarchy can be deleted.  Since UNIX doesn't have an atomic operation that
#   renames two files at a time, this method has a very small window where the
#   directory will not exist, but this is preferable to the many long-duration
#   inconsistencies that can occur with the other method.
#
#   Currently, this script can only mirror a directory at the top level of your
#   home directory (e.g. ~/www).  You also must have a ~/tmp directory (used as
#   a staging area) on both the master and slave machines.  ~/tmp is used rather
#   than /tmp because on some OSes that directory is mode 777 instead of 1777,
#   allowing people to delete your tar file before you get a chance to pull your
#   files out of it.  Also, the zcat picked by your $PATH on the remote machine
#   must be GNU's zcat.
#
#   This script works quite well as a cron job to mirror your directory at
#   regular intervals.  In this case, I suggest redirecting the script's
#   extensive stderr/stdout output to a log file so that if your mirroring ever
#   fails to occur you can determine why later.
#
#   Note that if you accidentally mirror to a machine that has the same NFS home
#   directory as the one you're logged into (or that _is_ the same machine
#   you're logged into), you will not lose your files (since deletion of the old
#   copy of the directory is not done until after the new one is in place).
#   However, if any files in the mirrored directory were hard links to files
#   outside the directory, they no longer will be after the mirror operation.
#
# DATE        MODIFICATION
# ==========  ==================================================================
# 1999-04-22  Original.

abort() {
    print "\n$mirror: Aborting at `date`."
    exit 1
}

execho() {
    print "\n% $@"
    $@
}

# Make sure we have gzip and whoami in our $PATH (cron default path usu. sucks).
PATH=/usr/local/bin:/opt/gnu/bin:/usr/ucb:$PATH 

mirror=${0##*/}  # get the root name of the script as installed, for printing

if [[ $# < 2 ]]; then
    print "Usage: $mirror <directory> [<user1>@]<host1> [[<user2>@]<host2>...]"
    exit 1
fi

print "$mirror: Started at `date`.\n"

dir=$1
shift
print "Directory: $dir"

trap 'execho rm -f ~/tmp/$dir.tar?(.gz)' 0 1 2 3 15

umask 77  # make sure people can't get at your otherwise protected files via tar

execho tar cvf ~/tmp/$dir.tar $dir || abort
execho gzip ~/tmp/$dir.tar || abort

while [[ $# > 0 ]]; do
    user_host=$1
    shift
    print
    if [[ $user_host = *@* ]]; then
	user=${user_host%%@*}
	host=${user_host##*@}
	print "Host: $host"
	print "User: $user"
	host_user="$host -l $user"
    else
	print "Host: $user_host"
	print "User: `whoami`"  # $USER not set under cron
	host_user=$user_host
    fi

    execho rcp -p ~/tmp/$dir.tar.gz $user_host:tmp || continue

    # Need to merge stderr with stdout first or ksh -x's trace output on stderr
    # will be out-of-sync with the other commands' (principally tar's) stdout.
    execho remsh $host_user "ksh -x 'exec 2>&1 && cd tmp"\
                                    "&& zcat $dir.tar.gz | tar xvf -"\
                                    "&& cd && mv $dir $dir.prev"\
                                    "&& mv tmp/$dir .'"

    execho remsh $host_user "rm -rf $dir.prev"
done

print "\n$mirror: Done at `date`."