rsunk

#!/usr/bin/env bash
#
#  Synopsis:
#	Archive /d1/d2/... to rsync.net and notify the masses via email.
#  Description:
#  	Rsync a precious directory (not fs) to an rsync.net server and send
#  	status email.  Files are deleted on remote, as well as excluded files.
#	dir /d1/d2/... sunk to
#
#		<rsyc.net>:sunk/host/$(hostname -f)/root/d1/d2/...
#
#	Multiple exclude files and emails to notify can be specified.
#
#	Email a summary report to addresses specified in --notify options.
#  Usage:
#	rsunk								\
#		--ignore-24						\
#		--notify jmscott@setspace.com				\
#		--notify jmscott@rj2tech.com				\
#		--ssh-auth 8008@usw-s008.rsync.net			\
#		--source-path /usr/local				\
#			>>log/rsunk-root.log
#	rsunk --notify jmscott --source-path /home/svn --ssh-auth jmscott
#	rsunk --notify jmscott --source-path /home/jmscott/backup 	\
#		--ssh-auth jmscott --itemize
#  Options:
#	--subject-tag		#  tag desccribint
#	--source-path <dir>	#  Path to root of source directory
#	--ssh-auth		#  Authorization at rsync.net
#	--dry-run		#  Do a dry run (no transfer)
#	--exclude-path <path>	#  path to exclude file.  multiple ok
#	--ignore-24		#  Ignore rsync exit status 24 (file gone)
#	--itemize		#  Detailed itemization of file changes
#	--notify <email>	#  Send summary report to <email>.  Multiple ok.
#	--one-file-system	#  Only one file system
#  See:
#	https://github.com/jmscott/work/rsunk
#	https://github.com/jmscott/work/rsync-fs
#	https://rsync.net
#  Note:
#	Add --fault-path or --logger for recording or sending faults to syslog!
#
#	Add option --help
#
#	Consider a --subject-tag option used in subject line of emailed report.
#	"Subject: OK: rsunk: nightly" or "ERROR: rsunk: rj2imac-local"
#

START_EPOCH=$(date +%s)
PROG=$(basename $0)
SOURCE_PATH=
TARGET=
SSH_AUTH=
COMPRESS=
DU_OUT=
DRY_RUN=
EXCLUDE_PATHS=

#  ignore exit status 24 (some files changed)
IGNORE_24=no

RSYNC_ARGS="
	--archive
	--compress
	--delete
	--delete-excluded
	--human-readable
	--itemize-changes
	--numeric-ids
	--prune-empty-dirs
	--relative
	--verbose
"

HOST=$(hostname -f)
TARGET_PATH=sunk/host/$HOST/root/
RSYNC_OUT=
RSYNC_TEE=cat
ITEMIZE=no

log() {
	echo "$(date +'%Y/%m/%d %H:%M:%S'): $PROG: $@"
}

leave() {
	rm -f $RSYNC_OUT
	log "good bye, cruel world: elapsed $(elapsed-english $START_EPOCH)"
}

catch() {
	log "WARN: caught signal: $1"
}

do_mail() {
	SUBJECT="$1"

	(
		cat
		cat <<END

Begin Process:

	Elapsed: $(elapsed-english $START_EPOCH)
  Hostname FQDN: $HOST
           User: $USER
    Source Path: $SOURCE_PATH
    Target Path: $TARGET_PATH
         Target: $TARGET
     Disk Usage: $DU_OUT
Disk Usage Path: $DU_PATH
  Du Error Path: $DU_ERR
        Itemize: $ITEMIZE
    Script Path: $0
  Exclude Paths: $EXCLUDE_PATHS
     Start Time: $RSYNC_START_TIME
       End Time: $(date --rfc-3339=seconds)
       SSH Auth: $SSH_AUTH
     Rsync Args: $RSYNC_ARGS
 Temp Rsync Out: $RSYNC_OUT
 Temp Rsync Tee: $RSYNC_TEE
 Ignore Exit 24: $IGNORE_24
    Which rsunk: $(which rsunk)
           PATH: $PATH
END
	#
	#  Concatenate rsync output. 
	# 
	if [ -s "$RSYNC_OUT" ];  then
log "WTF: '$RSYNC_OUT'"
		cat <<END

	Begin rsync stdout/err: $RSYNC_OUT

END
		cat $RSYNC_OUT
		cat <<END

	  End rsync stdout/err: $RSYNC_OUT
END
	fi

	#
	#  Process environment
	#
	cat <<END

	Begin Environment:
END
		env | sort
		cat <<END
	  End Environment:

END Process:
END
	) | tr -d '\r' | mail -s "$SUBJECT" $NOTIFY
}

#  die without mailing to users.
croak() {
	echo "ERROR: $@" >&2
	exit 1
}

#  die parsing command line options
eopt() {
	OPT=$1
	shift
	croak "option --$OPT: $@" 
}

eexclude() {
	eopt exclude-path $@
}

etwice() {
	eopt $1 'given more than one'
}

essh() {
	eopt ssh-auth $@
}

esrc() {
	eopt source-path $@
}

die() {
	log "ERROR: $@" >&2

	#
	#
	#
	MSG="$@"
	if [ -n "$SOURCE_PATH" ];  then
		MSG=$(basename $SOURCE_PATH)
	fi
	do_mail "ERROR: $PROG: $MSG" <<END
In process $PROG#$$ the following error occured:

    $@

For more details, see the output of standard error of this script log file,
probably defined in a crontab for the user $USER.

END
	exit 1
}

#  parse command line args

while [ "$1" ];  do
	ARG="$1"
	shift
	case "$ARG" in
	--exclude-path)
		EP="$1"

		test -z "$EP" && eexclude 'empty path to exclude list'
		test -r $EP || eexclude "can not read path: $EP"
		RSYNC_ARGS="$RSYNC_ARGS --exclude-from=$EP"
		EXCLUDE_PATHS="$EXCLUDE_PATHS $EP"
		shift
		;;
	--notify)
		N="$1"
		test -n "$N" || die 'option --notify: missing user@host'
		NOTIFY="$NOTIFY $N"
		shift
		;;
	--ssh-auth)				# user@host
		test -z "$SSH_AUTH" || essh 'given more than once'
		test -n "$1" || essh 'missing auth'

		#  Note: how to test if ssh auth is valid in config?

		SSH_AUTH="$1"
		shift
		;;
	--source-path)		#  file path
		test -z "$SOURCE_PATH" || etwice source-path

		SP="$1"

		#  frisk the source path
		case "$SP" in
		'')
			esrc 'missing path to source: /d1/d2/d3...'
			;;
		/)
			;;
		/*)
			test -e || esrc "directory does not exist: $1" 
			test -d || esrc "not a directory: $1"
			;;
		*)
			esrc "not full path: $SP"
			;;
		esac
		SOURCE_PATH="$SP"
		shift
		;;
	--one-file-system)
		test -z "$ONE_FS" || etwice one-file-system 
		RSYNC_ARGS="$RSYNC_ARGS --one-file-system"
		ONE_FS=seen
		;;
	--dry-run)
		RSYNC_ARGS="$RSYNC_ARGS --dry-run"
		DRY_RUN=yes
		;;
	--ignore-24)
		test $IGNORE_24 = no || etwice ignore-24
		IGNORE_24=yes
		;;
	--itemize)
		# itemize defaults to "no"
		test $ITEMIZE = yes && etwice itemize

		ITEMIZE=yes
		RSYNC_OUT=${TMPDIR:=/tmp}/$PROG.$$
		RSYNC_TEE="tee $RSYNC_OUT"
		;;
	--*)
		croak "unknown option: $ARG"
		;;
	*)
		croak "unknown cli argument: $ARG"
		;;
	esac
done
test -n "$SSH_AUTH" || croak 'missing option --ssh-auth'
test -n "$NOTIFY" || croak 'missing option --notify'
test -n "$SOURCE_PATH" || croak 'missing option: --source-path'

trap leave EXIT
trap 'catch TERM' SIGTERM
trap 'catch INT' SIGINT
trap 'catch QUIT' SIGQUIT

log 'hello, world'
log "host: $HOST"
log "PATH: $PATH"
log "source path: $SOURCE_PATH"

EXCLUDE_PATHS=$(echo "$EXCLUDE_PATHS" | sed 's/^  *//')
log "exclude paths: $EXCLUDE_PATHS"

NOTIFY=$(echo "$NOTIFY" | sed 's/^  *//')
log "notify: $NOTIFY"

log "start epoch: $START_EPOCH"
log "ssh auth: $SSH_AUTH"
log "ignore exit status 24: $IGNORE_24"

log "itemize: $ITEMIZE"

log "target path: $TARGET_PATH"

TARGET="$SSH_AUTH:$TARGET_PATH"
log "rsync target: $TARGET"
log 'starting rsync ...'

RSYNC_START_TIME=$(date --rfc-3339=seconds)
log "mkdir target path @rsync.net: $TARGET_PATH"
ssh -n $SSH_AUTH "mkdir -p $TARGET_PATH" ||
			die "mkdir target path failed: exit status=$?"

#
#  Call rsync.
#
log 'calling rsync ...'
time rsync 								\
	$RSYNC_ARGS							\
	$SOURCE_PATH $TARGET 2>&1					|
	$RSYNC_TEE
STATUS="${PIPESTATUS[*]}"
case $STATUS in
'0 0')
	log 'done, no errors in rsync'
	;;
'24 0')
	if [ $IGNORE_24 = yes ];  then
		log "WARN: (OK): some files changed during transfer"
	else
		die "rsync: some files changed during transfer"
	fi
	;;
*)
	die "rsync failed: exit status=$STATUS"
	;;
esac

#  only probe rsync.net for disk size if not a dry run
if [ "$DRY_RUN" != yes ];  then

	#  strip /+ pattern from path do to clumsy path construction
	DU_PATH=$(echo "$TARGET_PATH/$SOURCE_PATH" | tr --squeeze-repeats /)
	log "remote du path: $DU_PATH"

	DU_OUT=$(
		ssh -n $SSH_AUTH du -sh $DU_PATH 2>&1 </dev/null	|
		awk '{print $1, $2}'					|
		sed "s@/sunk/host/$HOST/@@"
	)
	case "$DU_OUT" in
	*[0-9]*[KMGT]*)
		log "disk usage: $DU_OUT"
		;;
	'')
		log 'empty du output'
		DU_ERR=$(
			realpath					\
				--canonicalize-missing			\
				--relative-base	sunk/host/$HOST/root	\
				$DU_PATH
		)
		die "du failed path: $DU_ERR"
		;;
	*)
		die "wierd du output: $DU_OUT"
		;;
	esac
fi

#  Trim up the Subject: line
SUBJECT=$(
	echo OK: $DU_OUT: $(elapsed-english $START_EPOCH)		|
	sed								\
		--expression '
			s/^[[:space:]]*//;
			s/[[:space:]]*$//;
			s/[[:space:]]+/ /g
		'							\
		--regexp-extended
)
		
echo 'Done. No Errors.'	| do_mail "$SUBJECT"
exit 0