From 1c561dc5e324cf82dcf70a91b681967b2fd35874 Mon Sep 17 00:00:00 2001
From: Max Christian Pohle
Date: Mon, 19 Apr 2021 01:19:53 +0200
Subject: Improved version of the squash script

---
 zfs-squash-datasets.sh | 191 ++++++++++++++++++-------------------------------
 1 file changed, 68 insertions(+), 123 deletions(-)

diff --git a/zfs-squash-datasets.sh b/zfs-squash-datasets.sh
index 3623edf..0e6d602 100755
--- a/zfs-squash-datasets.sh
+++ b/zfs-squash-datasets.sh
@@ -1,139 +1,84 @@
-#!/usr/local/bin/bash
-
-# DO NOT USE THIS SCRIPT
-# It is in a state of hot garbage and will probably be deleted soon
-
-
-print_help_and_exit() {
-    echo "Please run this program with a source and a target dataset name as arguments"
-    echo "e.g. $0 zpool/dataset1 zpool/dataset2"
-    exit
+#!/bin/sh
+# This script expects a source dataset and a target dataset as arguments.  It
+# will then copy each snapshot of the source dataset into the target as a
+# subfolder with the name of the dataset. This is being archived by cloning the
+# source-dataset first into a sub-dataset of the source dataset named after the
+# snapshot. From there rsync is used to copy files into the target location.
+# After that finishes a snapshot will be created in the target with the original
+# snapshot-name and a suffix, which is the last part of the source dataset.
+#
+# Known Bugs:
+#
+# * This script will immidiatly stop when any command returns an error and it
+# will not clean up automatically the clones it created. But these clones are
+# not promoted by the script and can be deleted without altering the source
+# dataset and in order to re-execute the script after fixing the reason for the
+# error.
+#
+# * There is currently no sanity checking. Please do not copy files from a
+# dataset, which is a child of the source dataset.
+
+#
+# Example:
+#
+# 	./zfs-squash-datasets.sh  zpool/software zpool/projects
+#
+# would create a folder /zpool/projects/software within the dataset
+# zpool/projects
+
+
+
+set -o errexit
+
+getMountPoint() {
+	zfs get -H -o value mountpoint "$1" ||\
+		(echo "mountpoint for dataset '$1' not found" ; exit 1)
 }
 
-SIMULATION=0
-
-args=$(getopt n $*)
-test $? == 0 || print_help_and_exit
-
-set -- $args
-
-while :; do
-    sleep 1
-    echo $1
-    case "$1" in
-        --) shift; break ;;
-        -n) SIMULATION=1 ; shift ;;
-    esac
-done
+listAllSnapshots() {
+	zfs list -H -o name -tsnap -r "$1" ||\
+		(echo "Could not find snapshots under $1" ; exit 2)
+}
 
-DATASET_ROOT=$1
+DATASET=$1
+DATASET_NAME=`basename $DATASET`
 DATASET_TARGET=$2
+TARGET_PATH=`basename $DATASET_TARGET`
 
-if ! zfs get mountpoint $DATASET_TARGET >/dev/null; then
-    echo "Dataset $DATASET_TARGET does not exist and will be created."
-    echo "You have 10 seconds to abort this with CTRL-C"
-    sleep 10
-    if ! $SIMULATION; then
-        zfs create $DATASET_TARGET
-    fi
-fi
-
-if zfs get -H -t filesystem -o name,value -r snapdir zeus/data/projects | grep -v visible >/dev/null; then
-    echo Please make all snapdirs visible:
-    zfs get -H -t filesystem -o name,value -r snapdir zeus/data/projects
-    exit 2
-fi
-
-
-MOUNTPOINT_TARGET=$(zfs get -o value -H mountpoint $DATASET_TARGET)
-test ${SIMULATION:0} = 1 && test $MOUNTPOINT_TARGET == "" && MOUNTPOINT_TARGET="datapool/simulation"
-
-test -d "$MOUNTPOINT_TARGET" || print_help_and_exit
+MP_TARGET=`getMountPoint $2`
 
-# make sure, that ctrl-c also works in the inner loops
-trap exit INT
+test -d "$MP_TARGET" || (echo "failed to find mountpoint for $DATASET_TARGET: Its not $MP_TARGET" ; exit 3)
 
-# recursively contains all snapshot names (the part after the @) from root on, sorted and only once
-SNAPSHOTS=$(zfs list -H  -oname -tsnap -r $DATASET_ROOT | cut -d@ -f2 | sort -u)
-DATASETS=$(zfs list -H -oname -t filesystem -r $DATASET_ROOT)
+for i in `listAllSnapshots $DATASET`; do
+	SNAPSHOT=`echo "$i" | cut -d@ -f2`
 
-test "${SIMULATION:-0}" = 1 && echo "SIMULATION MODE"
+	CLONETMP="$DATASET/$SNAPSHOT"
 
-$DATASETS
+	zfs clone "$i" "$CLONETMP" ||\
+		(echo "Failed to create clone" ; exit 4)
 
+	MP_CLONE=`getMountPoint $CLONETMP`
 
-echo "We are going to flatten the hierachy of this dataset:"
-printf "\t%s\n" $DATASETS
-echo -e "into the dataset with the name:"
-echo -e "\t$DATASET_TARGET"
-echo -e "mounted under:"
-echo -e "\t$MOUNTPOINT_TARGET"
-echo -e "with the following snapshots:"
-printf "\t%s\n" $SNAPSHOTS
-echo ""
-echo "You have 10 seconds to abort this with CTRL-C"
-sleep 10
+	test -d "$MP_CLONE" ||\
+		(echo "$CLONETMP: clones mount point not found" ; exit 5)
 
-set -- $SNAPSHOTS
-SNAPSHOTS_TOTAL=$#
-SNAPSHOT_CURRENT=0
 
-set -- $DATASETS
-shift
-SUBDATASETS=$@
+	# copy all files to the target location
+	(set -x ; rsync -ahox --info=progress2 --delete "$MP_CLONE/"  "$MP_TARGET/$TARGET_PATH" ||\
+		(echo "rsync exit status was $?" ; exit 6)
+	)
+	
+	(set -x ; zfs snap "$DATASET_TARGET@$SNAPSHOT-$DATASET_NAME" ||\
+		(echo "failed to create a snapshot $DATASET_TARGET: $SNAPSHOT-$DATASET_NAME" ; exit 7)
+	)
 
-for SNAPSHOT in $SNAPSHOTS; do
-    echo ""
-    echo ""
-    SNAPSHOT_CURRENT=$((SNAPSHOT_CURRENT + 1))
-    echo "[$SNAPSHOT_CURRENT/$SNAPSHOTS_TOTAL] Next snapshot is: $SNAPSHOT"
+	# move clone out of the source dataset, so that we can resume
+	# zfs promote "$CLONETMP" ||\
+	# 	(echo "Failed to promote clone '$CLONETMP'" ; exit 8)
+	zfs destroy "$CLONETMP" ||\
+		(echo "Failed to destroy clone '$CLONETMP'" ; exit 8)
 
-    for DATASET in $SUBDATASETS; do
-
-        DATASET_RELATIVE=${DATASET##$DATASET_ROOT}
-
-        MOUNTPOINT=$(zfs get -H -o value mountpoint $DATASET)
-        echo "RELATIVE=$DATASET_RELATIVE"
-
-        if [[ $MOUNTPOINT == "legacy" ]]; then
-            MOUNTPOINT=$(findmnt -n -o target $DATASET)
-            if test -d $MOUNTPOINT; then
-                >&2 echo "Dataset has a legacy mountpoint, but was not found in the fstab: $DATASET"
-                continue
-            fi
-        fi
-
-        if [[ $MOUNTPOINT == "-" ]]; then
-            >&2 echo "Dataset does not have a mount point or is a vdev: $DATASET"
-            continue
-        fi
-
-        SNAPDIR=$MOUNTPOINT/.zfs/snapshot/$SNAPSHOT
-        if test -d $SNAPDIR; then
-
-            TARGET_DIR="${MOUNTPOINT_TARGET}${DATASET_RELATIVE}"
-
-            echo -e \\trsync --info=stats3 --delete -a "$SNAPDIR/" "$TARGET_DIR/"
-            if test "${SIMULATION:-0}" = 0; then
-
-                mkdir -p "$TARGET_DIR"
-
-                if ! rsync --info=stats3 --delete -a "$SNAPDIR/" "$TARGET_DIR/" ; then
-                    >&2 echo "rsync could not entirely copy from $SNAPDIR to $DATASET_TARGET"
-                fi
-            fi
-        else
-            echo -e "\tSKIP: Snapshot $SNAPSHOT does not exist for dataset $DATASET"
-        fi
-    done
-
-    NEW_SNAPSHOT="$DATASET_TARGET@$SNAPSHOT"
-    echo ""
-    echo -e "\t> Concluding snapshot: ${NEW_SNAPSHOT} <"
-    echo ""
-
-    if test "${SIMULATION:-0}" = 0; then
-        zfs snap $NEW_SNAPSHOT
-    fi
+	# for better readability add two blank lines...
+	echo
+	echo
 done
-
-- 
cgit v1.2.3