zfs-backup/backup.sh

#!/usr/bin/env sh

nodes_file="nodes.txt"
### nodes.txt
# <origin host>   <user> <ssh port> <timezone> <compressed flag>
# <mirror 1 host> <user> <ssh port> <zfs root> <send speed>
# <mirror 2 host> <user> <ssh port> <zfs root> <send speed>
# ...

datasets_file="datasets.txt"
### datasets.txt
# <dataset 1> [remote dataset]
# <dataset 2> [remote dataset]
# <dataset 3> [remote dataset]
# ...
# recognized options: nosend nosnap(shot) remotezfs=<remote dataset name>

if [ ! -f $nodes_file ]; then echo "No nodes specified."; exit 2; fi
if [ ! -f $datasets_file ]; then echo "No datasets specified."; exit 3; fi

ZFS_SEND_OPTS="-pRLe"
ZFS_SNAPSHOT_OPTS="-r"
ZFS_RECV_OPTS="-vsu"
SSH_SEND_OPTS="-Ti ~/.ssh/id_rsa"
DEF_SPEED="20M"

read LOCAL_HOST LOCAL_USER LOCAL_PORT LOCAL_TZ SEND_COMPRESSED < $nodes_file

ORIGIN_PERMS="send,snapshot,hold"
MIRROR_PERMS="compression,receive,create,mount,mountpoint,readonly,jailed"

SEND_DATASETS=""
SNAPSHOT_DATASETS=""

debug()
{
    if [ "$BG_DEBUG_MODE" = "TRUE" ]; then
        echo $@
    fi
}

set_compression() {
    case $SEND_COMPRESSED in
        [tTyY]*) ZFS_SEND_OPTS="$ZFS_SEND_OPTS -c"
        debug "Enabled send as compressed datasets." ;;
    esac
}

populate_datasets()
{
    while read dataset dataset_options; do
        case $dataset in
            \#*) continue;;
        esac

        snapshot_dataset="yes"
        send_dataset="yes"
        for option in $dataset_options; do
            case $option in 
                "nosnap"*    )
                    snapshot_dataset=""
                    ;;
                "nosend"*    )
                    send_dataset=""
                    ;;
                "remotezfs="*)
                    remotezfs=`echo $option|cut -d'=' -f2`
                    dataset="$dataset:$remotezfs"
                    ;;
            esac
        done

        if [ $send_dataset ]; then
            SEND_DATASETS="$SEND_DATASETS $dataset"
            debug "Enabled send for $dataset."
        fi
        if [ $snapshot_dataset ]; then
            SNAPSHOT_DATASETS="$SNAPSHOT_DATASETS $dataset"
            debug "Enabled snapshot for $dataset."
        fi
    done < datasets.txt
}

list_datasets()
{
    for dataset in $SEND_DATASETS; do
        dataset=`echo $dataset|cut -d: -f1`
        zfs list -H $dataset
    done
}

list_remote_datasets() {
    while read addr user ssh_port zroot speed; do
        echo "=== $user@$addr $zroot ==="
        if [ $zroot ]; then
            case $zroot in 
                *"/") ;;
                *) zroot="$zroot/" ;;
            esac
        fi
        run_cmd="
            sh -c 'for dataset in $SEND_DATASETS; do
                dataset=`echo \\$dataset|cut -d: -f2`
                zfs list -H $zroot\$dataset;
            done'"
        if [ $BG_TEST_MODE = "TRUE" ]; then
            echo ssh $SSH_SEND_OPTS -o port=$ssh_port $user@$addr $run_cmd
        else
            ssh $SSH_SEND_OPTS -o port=$ssh_port $user@$addr $run_cmd
        fi
    done < sendnodes.tmp
}

get_latest_snapshot()
{
    zfs list -t snapshot $dataset | tail -n1 | cut -d'@' -f2 | cut -d' ' -f 1
}

get_latest_remote_snapshot()
{
    if [ $zroot ]; then
        case $zroot in 
            *"/") ;;
            *) zroot="$zroot/" ;;
        esac
    fi
    ssh $SSH_SEND_OPTS -o port=$ssh_port $user@$addr \
        "zfs list -t snapshot $zroot$dataset    |
            tail -n1|cut -d'@' -f2|cut -d' ' -f 1"
}

zfs_allow_origin()
{
    for dataset in $SEND_DATASETS; do
        dataset=`echo $dataset|cut -d: -f1`
        if [ $BG_TEST_MODE = "TRUE" ]; then
            echo zfs allow -u $LOCAL_USER $ORIGIN_PERMS $dataset
        else
            zfs allow -u $LOCAL_USER $ORIGIN_PERMS $dataset
        fi
    done
}

zfs_allow_mirrors()
{
    while read addr user ssh_port zroot speed; do
        echo "=== $user@$addr $zroot ==="
        if [ $zroot ]; then
            case $zroot in 
                *"/") ;;
                *) zroot="$zroot/" ;;
            esac
        fi
        run_cmd="
            for dataset in $SEND_DATASETS; do
                dataset=`echo \\$dataset|cut -d: -f2`
                zfs allow -u $user $MIRROR_PERMS $zroot\$dataset
            done
        "
        if [ $BG_TEST_MODE = "TRUE" ]; then
            echo ssh $SSH_SEND_OPTS -o port=$ssh_port $user@$addr $run_cmd
                
        else
            ssh $SSH_SEND_OPTS -o port=$ssh_port $user@$addr $run_cmd
        fi
    done < sendnodes.tmp
}

create_snapshots()
{
    snapshot_id="`date -I`-`date +%s`-$LOCAL_TZ"
    for dataset in $SNAPSHOT_DATASETS; do
        if [ $BG_TEST_MODE = "TRUE" ]; then
            echo zfs snapshot $ZFS_SNAPSHOT_OPTS $dataset@$snapshot_id
        else    
	        zfs snapshot $ZFS_SNAPSHOT_OPTS $dataset@$snapshot_id
        fi
    done
}

send_latest()
{
    while read addr user ssh_port zroot speed; do
        echo "=== $user@$addr $zroot ==="
        if [ $zroot ]; then
            case $zroot in 
                *"/") ;;
                *) zroot="$zroot/" ;;
            esac
        fi
        for dataset in $SEND_DATASETS; do
            remote_dataset=`echo $dataset|cut -d: -f2`
            dataset=`echo $dataset|cut -d: -f1`
            debug "Processing $dataset (remote $remote_dataset)."
            origin_snapshot=`get_latest_snapshot $dataset`
            run_cmd="
                zfs send $ZFS_SEND_OPTS $dataset@$origin_snapshot |
                pv -qL ${speed:-$DEF_SPEED} |
                ssh $SSH_SEND_OPTS -o port=$ssh_port               \
                    $user@$addr                                    \
                    zfs recv $ZFS_RECV_OPTS $zroot$remote_dataset
                "
            if [ $BG_TEST_MODE = "TRUE" ]; then
                echo $run_cmd
            else
                $run_cmd
            fi
        done
    done < sendnodes.tmp
}

send_increment()
{
    while read addr user ssh_port zroot speed; do
        echo "=== $user@$addr $zroot ==="
        if [ $zroot ]; then
            case $zroot in 
                *"/") ;;
                *) zroot="$zroot/" ;;
            esac
        fi
        for dataset in $SEND_DATASETS; do
            remote_dataset=`echo $dataset|cut -d: -f2`
            dataset=`echo $dataset|cut -d: -f1`
            origin_snapshot=`get_latest_snapshot`
            remote_snapshot=`get_latest_remote_snapshot`
            run_cmd="
                zfs send $ZFS_SEND_OPTS          \
                    -i $dataset@$remote_snapshot \
                    $dataset@$origin_snapshot    | 
                pv -qL ${speed:-$DEF_SPEED} |
                ssh $SSH_SEND_OPTS -o port=$ssh_port $user@$addr     \
                        zfs recv $ZFS_RECV_OPTS $zroot$remote_dataset
            "
            if [ $BG_TEST_MODE = "TRUE" ]; then
                echo $run_cmd
            else
                $run_cmd
            fi
        done
    done < sendnodes.tmp
}


################################### MAIN ####################################

### Options
set -- `getopt "tdl:" "$@"` || {
    echo "Usage: `basename $0` [-tdl] [command]" 1>&2
    exit 1
}
BG_TEST_MODE=FALSE
while :; do
    case "$1" in
        -t)
            BG_TEST_MODE=TRUE
            echo Test mode.
        ;;
        -d)
            BG_DEBUG_MODE=TRUE
            ZFS_SEND_OPTS="$ZFS_SEND_OPTS -v"
            # Current on by default.
            # ZFS_RECV_OPTS="$ZFS_RECV_OPTS -v"
            echo Debug mode.
        ;;
        --) shift; break ;;
    esac
    shift
done

tail -n +2 $nodes_file|grep -v '^#' > sendnodes.tmp

set_compression
populate_datasets

### Command
case "$1" in
    list) list_datasets ;;
    list-remote) list_remote_datasets ;;
    snapshot) create_snapshots ;;
    configure-mirrors) zfs_allow_mirrors ;;
    configure-origin) zfs_allow_origin ;;
    mirror-reset) send_latest ;;
    mirror-increment) send_increment ;;
    #mirror-prune) send_increment_with_F ;;
esac

rm sendnodes.tmp