#!/bin/bash

# Author: Carl Baldwin & Alan Pippin
# Description: This script replicates a given zfs filesystem to a given zfs pool.
#              This script will keep all snapshots in sync, removing the ones
#              that have been deleted since the last replicate was performed.
#              This script will only send the new, or missing, snapshots since
#              the last replicate was performed.

# In test mode (test=1) commands are echoed, not executed
test=0

[ $test == 0 ] && exec >> /var/log/zfs/zfs-replicate.log 2>&1

# Usage: zfs-backup [filesystem] [destination_pool]
# This script has a limitation with children under a given filesystem.
# You must initially backup the parent filesystems first using this script
# before backing up any of the children filesystems.

fs=$1
fs=${fs%/}
fsname=${1#*/}
srcpool=${1%%/*}
srcfs="${srcpool}/$fsname"
srcfs=${srcfs%/}
dstpool=$2
dstfs="${dstpool}/$srcfs"
dstfs=${dstfs%/}
nodstsnaps=0
common=""

if [ $test == 1 ]; then
  echo "fs: $fs"
  echo "fsname: $fsname"
  echo "srcpool: $srcpool"
  echo "srcfs: $srcfs"
  echo "dstpool: $dstpool"
  echo "dstfs: $dstfs"
fi

if ! zpool list -H "$srcpool" >/dev/null 2>&1; then
  echo >&2 "-E- The source pool, '$srcpool' doesn't seem to exist."
  exit 1
fi

if ! zpool list -H "$dstpool" >/dev/null 2>&1; then
  echo >&2 "-E- The destination pool, '$dstpool' doesn't seem to exist."
  exit 1
fi

if ! zfs list -rH -t snapshot "$dstfs" 2>&1 | grep "$dstfs@" > /dev/null 2>&1; then
  echo >&2 "-W- No snapshots detected on the destination drive for this filesystem: $dstfs"
  if zfs list -t filesystem | grep "$dstfs"; then
   echo >&2 "-I- Found zfs filesystem $dstfs on the destination pool $dstpool without any snapshots"
   echo >&2 "-I- Removing the zfs filesystem: $dstfs"
   zfs destroy "$dstfs" 
  fi
  nodstsnaps=1
fi

if [ $nodstsnaps == 0 ]; then
  zfs list -rH -t snapshot $srcfs | grep "$srcfs@" | awk '{print $1}' > /tmp/source-list
  zfs list -rH -t snapshot $dstfs | grep "$dstfs@" | sed "s,$dstpool/,," | awk '{print $1}' > /tmp/destination-list
  diff -u /tmp/source-list /tmp/destination-list | grep -v '^+++' | awk '/^\+/ {print}' | sed "s,^\+,$dstpool/," > /tmp/obsolete-snapshots
  rm -f /tmp/source-list /tmp/destination-list

  echo >&2 "Removing obsolete backups from the destination pool" 
  for snapshot in $(cat /tmp/obsolete-snapshots); do
    echo >&2 "Removing '$snapshot' from destination."
    [ $test == 0 ] && zfs destroy "$snapshot"
  done

  echo >&2 "Rolling back to the most recent snapshot on the destination." 
  [ $test == 0 ] && zfs rollback $(zfs list -rH -t snapshot $dstfs | grep "$dstfs@" | awk '{snap=$1} END {print snap}')

  echo >&2 "Calculating the most recent common snapshot between the two filesystems." 
  if zfs list -H "$dstfs" > /dev/null 2>&1; then
    for snap in $(zfs list -rH -t snapshot "$dstfs" | grep "$dstfs@" |
                    sed 's,.*@,,' | awk '{print$1}'); do
      if zfs list -rH -t snapshot "$fs" | grep "$fs@" | sed 's,.*@,,' | awk '{print$1}' | grep "^${snap}$" >/dev/null 2>&1; then
        common=$snap
      fi  
    done
  fi
fi

base=$common
foundcommon=false
if [ -z "$common" ]; then
  foundcommon=true
fi

for snap in $(zfs list -rH -t snapshot "$fs" | grep "$fs@" |
                sed 's,.*@,,' | awk '{print$1}'); do
  if [ "$snap" = "$common" ]; then
    foundcommon=true
    continue
  fi

  if $foundcommon; then
    if [ -z "$base" ]; then
      echo >&2 "Sending '$fs@$snap'"
      [ $test == 0 ] && zfs set readonly=on "$dstpool"
      [ $test == 0 ] && zfs set atime=off "$dstpool"
      [ $test == 0 ] && zfs set sharenfs=off "$dstpool"
      [ $test == 0 ] && zfs set mountpoint=legacy "$dstpool"
      [ $test == 0 ] && zfs send "$fs@$snap" | zfs recv -v "$dstfs"
    else
      echo >&2 "Sending '$fs@$base' -> '$fs@$snap'"
      [ $test == 0 ] && zfs send -i "$fs@$base" "$fs@$snap" | zfs recv -v "$dstfs"
    fi
    base=$snap
  fi
done

true