From bcec7a9c3f4c857b292657e00753c74f1d789cc3 Mon Sep 17 00:00:00 2001
From: Alice Gaudon <alice@gaudon.pro>
Date: Sat, 13 Mar 2021 14:37:39 +0100
Subject: [PATCH] zsc-sync.sh: sync subdatasets in init if they are out of date
 with their parents

---
 zsc-sync.sh | 80 +++++++++++++++++++++++++++++++++++++++++------------
 1 file changed, 63 insertions(+), 17 deletions(-)

diff --git a/zsc-sync.sh b/zsc-sync.sh
index 517dd64..127badd 100755
--- a/zsc-sync.sh
+++ b/zsc-sync.sh
@@ -56,12 +56,8 @@ resume_transfer() {
 handle_dataset() {
 	DATASET=$1
 	RECURSIVE=$2
-	D=$3
-	W=$4
-	M=$5
-	Y=$6
 
-	PREFIX=$7
+	PREFIX=$3
 	if test -z "$PREFIX"; then
 		PREFIX="zsc-"
 	fi
@@ -70,10 +66,13 @@ handle_dataset() {
 		PREFIX=""
 	fi
 
-	TARGET_REMOTE_DATASET="$REMOTE_ROOT_DATASET/"$(echo -e "$DATASET" | sed -r "s/\//_/g")
+	TARGET_REMOTE_DATASET=$4
+	if test -z "$TARGET_REMOTE_DATASET"; then
+		TARGET_REMOTE_DATASET="$REMOTE_ROOT_DATASET/"$(echo -e "$DATASET" | sed -r "s/\//_/g")
+	fi
 
 	echo "[$DATASET -> $TARGET_REMOTE_DATASET]"
-	echo -e
+	echo
 
 
 	resume_transfer "$TARGET_REMOTE_DATASET" || return 0
@@ -85,7 +84,7 @@ handle_dataset() {
 
 	echo "Remote snapshots: ${#REMOTE_SNAPSHOTS[@]}"
 	echo "Local snapshots: ${#LOCAL_SNAPSHOTS[@]}"
-	echo -e
+	echo
 
 	for local_snapshot in "${LOCAL_SNAPSHOTS[@]}"; do
 		found=false
@@ -107,7 +106,7 @@ handle_dataset() {
 	done
 
 	echo "Snapshots are synchronized up to $FROM_SNAPSHOT"
-	echo -e
+	echo
 
 	if test ! -z "$TO_SNAPSHOT"; then
 		echo "Synchronizing snapshots from $FROM_SNAPSHOT to $TO_SNAPSHOT ..."
@@ -120,13 +119,16 @@ handle_dataset() {
 		fi
 		zfs send -v -p $recu $incr "$DATASET@$TO_SNAPSHOT" | ssh $REMOTE zfs receive -F -s "$TARGET_REMOTE_DATASET" || return 1
 
+		echo
 		echo "Done"
+		echo
 	fi
 }
 
 init_subdataset() {
 	DATASET=$1
 	PARENT_DATASET=$2
+	LATEST_SYNCHRONIZED_SNAPSHOT=$3
 
 	TARGET_REMOTE_DATASET="$REMOTE_ROOT_DATASET/"$(echo -e $PARENT_DATASET | sed -r "s/\//_/g")"/"${DATASET:$(echo -e "$PARENT_DATASET" | wc -c)}
 
@@ -146,27 +148,71 @@ init_subdataset() {
 		echo "[$DATASET@$LOCAL_SNAPSHOT -> $TARGET_REMOTE_DATASET]"
 		zfs send -v -p "$DATASET@$LOCAL_SNAPSHOT" | ssh $REMOTE zfs receive -F -s "$TARGET_REMOTE_DATASET" || return 1
 
-		echo -e "Done"
+		echo
+		echo "--- Done"
+		echo
+	elif [ $(echo -e "$remote_snapshots" | grep -e "^$TARGET_REMOTE_DATASET@$LATEST_SYNCHRONIZED_SNAPSHOT\$" | wc -l) -eq 0 ]; then
+		echo
+		echo "--- Subdataset $TARGET_REMOTE_DATASET is out of date compared to parent dataset. Syncing first..."
+
+		handle_dataset "$DATASET" false "" "$TARGET_REMOTE_DATASET" || exit 1
+
+		echo
+		echo "--- Done"
+		echo
 	fi
 }
 
 
 for line in $(cat "$ZSC_CONFIG") ; do
 	IFS=: read dataset recursive d w m y prefix <<< "$line"
-	echo "| ##### Dataset: $dataset Recursive: $recursive Prefix: $prefix d: $d w: $w m: $m y: $y #####"
+	echo
+	echo
+	echo "##### Dataset: $dataset Recursive: $recursive Prefix: $prefix d: $d w: $w m: $m y: $y #####"
+	echo
 
 	if [ $recursive = "true" ]; then
 		SUBDATASETS=($(zfs list -o name | grep "$dataset/"))
-		echo -e "| Subdatasets: ${SUBDATASETS[@]}"
-		echo -e '|'
-		for p in "${SUBDATASETS[@]}"; do
-			init_subdataset $p $dataset | awk '{print "| " $0}' || exit 1
+
+
+		# Find latest synchronized parente snapshot
+		REMOTE_DATASET="$REMOTE_ROOT_DATASET/"$(echo -e "$dataset" | sed -r "s/\//_/g")
+		REMOTE_SNAPSHOTS=($(echo -e "$remote_snapshots" | grep "$REMOTE_DATASET@" | cut -d"@" -f 2))
+		LOCAL_SNAPSHOTS=($(zfs list -H -t snapshot -o name -s creation "$dataset" | cut -d"@" -f 2))
+
+		for local_snapshot in "${LOCAL_SNAPSHOTS[@]}"; do
+			found=false
+			for remote_snapshot in "${REMOTE_SNAPSHOTS[@]}"; do
+				if [ "$remote_snapshot" = "$local_snapshot" ] ; then
+					found=true
+					break
+				fi
+			done
+
+			if [ $found = true ] ; then
+				latest_synchronized_snapshot=$local_snapshot
+			fi
 		done
+		echo "Latest synchronized snapshot locally: $latest_synchronized_snapshot"
+		echo
+
+
+		# Init subdatasets
+		echo "Subdatasets: ${SUBDATASETS[@]}"
+		echo
+		for p in "${SUBDATASETS[@]}"; do
+			init_subdataset $p $dataset "$latest_synchronized_snapshot" || exit 1
+		done
+		echo
+		echo "----------"
+		echo
 	fi
 
-	handle_dataset $dataset $recursive $d $w $m $y $prefix | awk '{print "| " $0}' || exit 1
+	handle_dataset $dataset $recursive $prefix || exit 1
 
-	echo -e "----------------------------------------------------------------------"
+	echo
+	echo "----------------------------------------------------------------------"
+	echo
 done