bashclub-zfs-push-pull/backup-zfs

#!/bin/bash
# backup-zfs: use zfs send/recv to push/pull snapshots
prog="$(basename "$0")"

usage() {
	cat >&2 <<-EOF
	usage: $prog [-hvq] [-t tag] [-k keep] [-d dateopts] src dest
	  use zfs send/recv to push/pull snapshots

	  src          the source fs, specified as [host:]pool/path/to/fs
	  dest         the destination fs parent, specified as [host:]pool/path/to/fs
	               (the final path component of src will be appended to dest)
	  -p           ssh port
	  -h           help
	  -v           verbose mode
	  -q           quiet mode
	  -t tag       tag to use for naming snapshots (default: backup-zfs)
	  -k keep      number of snapshots to keep on src (default: 5)
	  -d dateopts  options for date(1) - used to name the snapshots (default: +%F_%T)
	  -s           store mode - output snaps from local fs to ssh server
	  -r           read mode - read snaps from ssh server to local fs
	  -g gpg-id    gpg recipient key id (store mode only)
	EOF
	exit $1
}

# log to syslog; if verbose or on a tty, also to stdout
# usage: log msg
log() {
	logger -t "$prog" -- "$@"
	if ! $quiet && [[ -t 1 ]] || $verbose ; then
		echo "$@" >&2
	fi
}

# exit with a code & message
# usage: die $exitcode msg
die() {
	code="$1"
	shift
	if [[ $code -ne 0 ]] ; then
		verbose=true log "FATAL: $@"
	else
		log "$@"
	fi
	exit $code
}

# run zfs(1) command either locally or via ssh
# usage: ZFS "$host" command args...
ZFS() {
	host="$1"
	shift
	if [[ -n $host ]] ; then
		log "remote ($host): zfs $@"
		ssh -C -p $port "$host" zfs "$@"
	else
		log "local: zfs $@"
		zfs "$@"
	fi
}

###
### defaults
###
tag="$prog"
dateopts="+%F_%T"
keep=5
verbose=false
quiet=false
tossh=false
fromssh=false
port=22

###
### parse options
###
while getopts "hvqk:p:t:d:srg:" opt ; do
	case $opt in
		h) usage 0 ;;
		v)
			verbose=true
			send_opts="-v"
			recv_opts="-v"
			;;
		q) quiet=true ;;
		k) keep=$OPTARG ;;
		p) port=$OPTARG ;;
		t) tag=$OPTARG ;;
		d) dateopts=$OPTARG ;;
		s) tossh=true ;;
		r) fromssh=true ;;
		g) gpgid="$OPTARG" ;;
		*) usage 1 ;;
	esac
done
shift $((OPTIND-1))
date="$(date $dateopts)"
$tossh && $fromssh && die 1 "-s and -r are mutually exclusive"
if ! $tossh && [[ -n $gpgid ]] ; then
	die 1 "-g can only be used with -s"
fi

###
### parse src & dest host/fs info
###
# fail if there's ever >1 colon
if [[ $1 =~ :.*: || $2 =~ :.*: ]] ; then
	die 1 "invalid fsspec: '$1' or '$2'"
fi

# fail if src or dest isn't specified
if [[ -z $1 || -z $2 ]] ; then
	usage 1
fi
src="$1"
dest="$2"

###
### ssh mode - output snaps from local fs to ssh or read snaps from ssh to local fs
if $tossh ; then
	log "sending from local zfs filesystem to SSH server"

	# make sure src exists
	if [[ $src =~ : ]] ; then
		die 1 "$src must be a local zfs filesystem"
	elif [[ $(ZFS "" list -H -o name "$src" 2>/dev/null) != $src ]] ; then
		die 1 "$src must be a local zfs filesystem"
	fi

	# split dest to components
	if [[ $dest =~ : ]] ; then
		desthost="${dest%:*}"
		destpath="${dest#*:}"
	else
		die 1 "$dest must be ssh host:path"
	fi

	# get the last src component
	srcbase="${src##*/}"

	###
	### create new snapshot on src
	###
	snap="${tag}_$date"
	cur="$src@$snap"
	ZFS "$srchost" snapshot -r "$cur" || die $? "zfs snapshot failed"

	###
	### get newest snapshot on dest - it must exist on src
	###
	#last="$(ZFS "$desthost" list -d 1 -t snapshot -H -S creation -o name $destfs/$srcbase | head -n1 | cut -f2 -d@)"
	last="$(ssh "$desthost" zfslast)"

	###
	### send
	###
	# refuse to send without a valid .last maker
	if [[ -z $last ]] ; then
		die 1 "ssh path contains no .last file"
	# special case: tagged snapshots exist on dest, but src has rotated through all
	elif ! ZFS "$srchost" list $src@$last &>/dev/null ; then
		die 1 "no incremental path from from $src to $dest"
	# normal case: send incremental
	else
		log "sending $([[ -n $gpgid ]] && echo "encrypted ")incremental snapshot from $src to $dest (${last#${tag}_}..${cur#*@${tag}_})"
		#ZFS "$srchost" send $send_opts -R -i "$last" "$cur" | ZFS "$desthost" receive $recv_opts -Fue "$destfs" || die $? "zfs incremental send failed"
		if [[ -n $gpgid ]] ; then
			ZFS "$srchost" send $send_opts -R -i "$last" "$cur" \
				| gpg --trust-model always --encrypt --recipient "$gpgid" \
				| ssh "$desthost" zfswrite "${tag}_$date.zfssnap.gpg" \
				|| die $? "zfs incremental send failed"
			ssh "$desthost" zfslast "$snap"
		else
			ZFS "$srchost" send $send_opts -R -i "$last" "$cur" \
				| ssh "$desthost" zfswrite "${tag}_$date.zfssnap" \
				|| die $? "zfs incremental send failed"
			ssh "$desthost" zfslast "$snap"
		fi
	fi

	exit
elif $fromssh ; then
	log "receving from SSH server to local zfs filesystem"

	# make sure dest exists
	if [[ $dest =~ : ]] ; then
		die 1 "$dest must be a local zfs filesystem"
	elif [[ $(ZFS "" list -H -o name "$dest" 2>/dev/null) != $dest ]] ; then
		die 1 "$dest must be a local zfs filesystem"
	fi

	# split src into components
	if [[ $src =~ : ]] ; then
		srchost="${src%:*}"
		srcpath="${src#*:}"
	else
		die 1 "$src must be ssh host:path"
	fi

	###
	### receive
	###
	log "receiving incremental snapshot from $src to $dest"
	#ZFS "$srchost" send $send_opts -R -i "$last" "$cur" | ZFS "$desthost" receive $recv_opts -Fue "$destfs" || die $? "zfs incremental send failed"
	for file in $(ssh "$srchost" zfsfind | sort) ; do
		log "receiving $file from $srchost"
		if [[ $file =~ \.gpg$ ]] ; then
			ssh "$srchost" zfsget "$file" | gpg | ZFS "$desthost" receive $recv_opts -Fue "$dest" \
				&& ssh "$srchost" rm "$file"
		else
			ssh "$srchost" zfsget "$file" | ZFS "$desthost" receive $recv_opts -Fue "$dest" \
				&& ssh "$srchost" rm "$file"
		fi
	done

	exit
fi

# discard anything before a colon to get the fs
srcfs="${src#*:}"
destfs="${dest#*:}"

# iff there is a colon, discard everything after it to get the host
[[ $src =~ : ]] && srchost="${src%:*}"
[[ $dest =~ : ]] && desthost="${dest%:*}"

# get the last src component
srcbase="${srcfs##*/}"

# ensure the destination fs exists before proceeding
if [[ $(ZFS "$desthost" list -H -o name "$destfs" 2>/dev/null) != $destfs ]] ; then
	die 1 "destination fs '$destfs' doesn't exist"
fi

###
### create new snapshot on src
###
cur="$srcfs@${tag}_$date"
ZFS "$srchost" snapshot -r "$cur" || die $? "zfs snapshot failed"

###
### get newest snapshot on dest - it must exist on src
###
last="$(ZFS "$desthost" list -d 1 -t snapshot -H -S creation -o name $destfs/$srcbase | head -n1 | cut -f2 -d@)"

###
### send & receive
###
# 1st time: send full snapshot
if [[ -z $last ]] ; then
	log "sending full recursive snapshot from $src to $dest"
	ZFS "$srchost" send $send_opts -R "$cur" | ZFS "$desthost" receive $recv_opts -Fue "$destfs" || die $? "zfs full send failed"
# special case: tagged snapshots exist on dest, but src has rotated through all
elif ! ZFS "$srchost" list $srcfs@$last &>/dev/null ; then
	die 1 "no incremental path from from $src to $dest"
# normal case: send incremental
else
	log "sending incremental snapshot from $src to $dest (${last#${tag}_}..${cur#*@${tag}_})"
	ZFS "$srchost" send $send_opts -R -i "$last" "$cur" | ZFS "$desthost" receive $recv_opts -Fue "$destfs" || die $? "zfs incremental send failed"
fi

###
### clean up old snapshots
###
for snap in $(ZFS "$srchost" list -d 1 -t snapshot -H -S creation -o name $srcfs \
	      | grep -F "@${tag}_" | cut -f2 -d@ | tail -n+$((keep+1)) ) ;
do
	ZFS "$srchost" destroy -r $srcfs@$snap
done