crossover/crossover

#!/bin/bash

# Cross Pool Migration and incremental replication Tool for Proxmox VMs using Ceph.
# Author: Bastian Mäuser <bma@netz.org>        

LC_ALL="en_US.UTF-8"

source rainbow.sh

# Predefine if you want
declare opt_influx_api_url=''
declare opt_influx_token=''
declare opt_influx_bucket=''
declare opt_influx_api_org=''
declare opt_influx_jobname=''
declare opt_influx_job_metrics='crossover_xmit'
declare opt_influx_summary_metrics='crossover_jobs'

name=$(basename "$0")
# readonly variables
declare -r NAME=$name
declare -r VERSION=0.8
declare -r PROGNAME=${NAME%.*}
declare -r PVE_DIR="/etc/pve"
declare -r PVE_NODES="$PVE_DIR/nodes"
declare -r QEMU='qemu-server'
declare -r QEMU_CONF_CLUSTER="$PVE_NODES/*/$QEMU"
declare -r EXT_CONF='.conf'
declare -r PVFORMAT_FULL='e:%t r:%e c:%r a:%a %b %p'
declare -r PVFORMAT_SNAP='e:%t c:%r a:%a %b'
logfile=$(mktemp)
declare -r LOG_FILE=$logfile

# associative global arrays 
declare -A -g pvnode
declare -A -g dstpvnode
declare -A -g svmids
declare -A -g dvmids

declare -g scluster
declare -g dcluster
declare -g scephversion
declare -g dcephversion

# global integers
declare -g -i perf_freeze_ok=0
declare -g -i perf_freeze_failed=0
declare -g -i perf_ss_ok=0
declare -g -i perf_ss_failed=0
declare -g -i perf_ss_ok=0
declare -g -i perf_ss_failed=0
declare -g -i perf_full_ok=0
declare -g -i perf_full_failed=0
declare -g -i perf_diff_ok=0
declare -g -i perf_diff_failed=0
declare -g -i perf_bytes_full=0
declare -g -i perf_bytes_diff=0
declare -g -i perf_bytes_total=0
declare -g -i perf_vm_running=0
declare -g -i perf_vm_stopped=0
declare -g -i perf_snaps_removed=0
declare -g -i perf_vm_total=0
declare -g -i perf_vm_ok=0

# commandline parameters
declare opt_destination
declare opt_vm_ids=''
declare opt_snapshot_prefix='mirror-'
declare opt_rewrite=''
declare opt_pool='rbd'
declare -i opt_prefix_id
declare opt_exclude_vmids=''
declare -i opt_debug=0
declare -i opt_dry_run=0
declare -i opt_syslog=0
declare -i opt_lock=1
declare -i opt_keepslock=0
declare -i opt_keepdlock=0
declare -i opt_overwrite=0
declare -i opt_online=0
declare -i opt_migrate=0
declare -i opt_noconfirm=0
declare opt_keep_local='0s'
declare opt_keep_remote='0s'


declare -r redstconf='^\/etc\/pve\/nodes\/(.*)\/qemu-server\/([0-9]+).conf$'
declare -r recephimg='([a-zA-Z0-9]+)\:(.*)'
declare -r restripsnapshots='/^$/,$d'
declare -r redateex='^([0-9]{4})([0-9]{2})([0-9]{2})([0-9]{2})([0-9]{2})([0-9]{2})$'
declare -r restripansicolor='s/\x1b\[[0-9;]*m//g'

function usage(){
   shift

    if [ "$1" != "--no-logo" ]; then
        cat << EOF
 _____                                 
|     |___ ___ ___ ___ ___ _ _ ___ ___ 
|   --|  _| . |_ -|_ -| . | | | -_|  _|
|_____|_| |___|___|___|___|\_/|___|_|   v$VERSION

EOF
    fi

    cat << EOF
Mirror tool Proxmox VMs on Ceph storages

Usage:
    $PROGNAME <COMMAND> [ARGS] [OPTIONS]
    $PROGNAME help
    $PROGNAME version
 
    $PROGNAME mirror     --vmid=<string> --destination=<destionationhost> --pool=<targetpool> --keeplocal=n --keepremote=n
Commands:
    version              Show version program
    help                 Show help program
    mirror               Replicate a stopped VM to another Cluster (full clone)

Options:
    --vmid               The source+target ID of the VM/CT, comma separated (eg. --vmid=100:100,101:101), or all for all
    --prefixid           Prefix for VMID's on target System [optional]
    --excludevmids       Exclusde VM IDs when using --vmid==all
    --destination        Target PVE Host in target pool. e.g. --destination=pve04
    --pool               Ceph pool name in target pool. e.g. --pool=data [default=rbd]
    --keeplocal          How many additional Snapshots to keep locally. e.g. --keeplocal=2d
    --keepremote         How many additional Snapshots to keep remote. e.g. --keepremote=7d
    --rewrite            PCRE Regex to rewrite the Config Files (eg. --rewrite='s/(net0:)(.*)tag=([0-9]+)/\1\2tag=1/g' would
                         change the VLAN tag from 5 to 1 for net0.
    --influxurl          Influx API url (e.g. --influxurl=https://your-influxserver.com/api/)
    --influxtoken        Influx API token with write permission
    --influxbucket       Influx Bucket to write to (e.g. --influxbucket=telegraf/autogen)
    --jobname            Descriptive name for the job, used in Statistics
    --mail               Mail address to send report to, comma-seperated (e.g. --mail=admin@test.com,admin2@test.com)
Switches:
    --online             Allow online Copy
    --migrate            Stop VM on Source Cluster before final Transfer and start on destination Cluster
    --nolock             Don't lock source VM on Transfer (mainly for test purposes)
    --keep-slock         Keep source VM locked on Transfer
    --keep-dlock         Keep VM locked after transfer on Destination
    --overwrite          Overwrite Destination
    --noconfirm          Don't ask for confirmation before starting --migrate mode (use with care!)
    --debug              Show Debug Output

Report bugs to <mephisto@mephis.to>

EOF
}

function parse_opts(){
#    local action=$1
    shift

    local args
    args=$(getopt \
           --options '' \
           --longoptions=vmid:,prefixid:,excludevmids:,destination:,pool:,keeplocal:,keepremote:,rewrite:,influxurl:,influxorg:,influxtoken:,influxbucket:,jobname:,mail:,online,migrate,nolock,keep-slock,keep-dlock,overwrite,dry-run,noconfirm,debug,syslog \
           --name "$PROGNAME" \
           -- "$@") \
           || end_process 128

    eval set -- "$args"

    while true; do    
      case "$1" in
        --vmid) opt_vm_ids=$2; shift 2;;
        --prefixid) opt_prefix_id=$2; shift 2;;
        --excludevmids) opt_exclude_vmids=$2; shift 2;;
        --destination) opt_destination=$2; shift 2;;
        --pool) opt_pool=$2; shift 2;;
        --keeplocal) opt_keep_local=$2; shift 2;;
        --keepremote) opt_keep_remote=$2; shift 2;;
        --rewrite) opt_rewrite=$2; shift 2;;
        --influxurl) opt_influx_api_url=$2; shift 2;;
        --influxorg) opt_influx_api_org=$2; shift 2;;
        --influxtoken) opt_influx_token=$2; shift 2;;
        --influxbucket) opt_influx_bucket=$2; shift 2;;
        --jobname) opt_influx_jobname=$2; shift 2;;
        --mail) opt_addr_mail="$2"; shift 2;;
        --online) opt_online=1; shift ;;
        --migrate) opt_migrate=1; shift ;;
        --dry-run) opt_dry_run=1; shift;;
        --noconfirm) opt_noconfirm=1; shift;;
        --debug) opt_debug=1; shift;;
        --nolock) opt_lock=0; shift;;
        --keep-slock) opt_keepslock=1; shift;;
        --keep-dlock) opt_keepdlock=1; shift;;
        --overwrite) opt_overwrite=1; shift;;
        --syslog) opt_syslog=1; shift;;
        --) shift; break;;
        *) break;;
      esac
    done

    if [ $opt_debug -eq 1 ]; then
        log info "============================================"
        log info "Proxmox Crosspool Migration: $VERSION";
        log info "pid: $(cat /var/run/"$PROGNAME".pid)"
        log info "============================================"
        log info "Proxmox VE Version:"

        echowhite "$(pveversion)"

        log info "============================================"
    fi

    [ -z "$opt_vm_ids" ] && { log info "VM id is not set."; end_process 1; }

    [ -z "$opt_influx_jobname" ] && { log info "Jobname is not set."; end_process 1; }

    if [ -n "$opt_keep_local" ]; then
        if ! [[ ${opt_keep_local:(-1)} == "s" || ${opt_keep_local:(-1)} == "d" ]]; then
            echo "--keeplocal: Parameter malformed. suffix s or d missing"
            end_process 255
        fi
    fi

    if [ -n "$opt_keep_remote" ]; then
        if ! [[ ${opt_keep_remote:(-1)} == "s" || ${opt_keep_remote:(-1)} == "d" ]]; then
            echo "--keepremote: Parameter malformed. suffix s or d missing"
            end_process 255
        fi
    fi

    if [ $opt_keepdlock -eq 1 ] && [ $opt_migrate -eq 1 ]; then
        log error "--keepdlock/--migrate: Invalid parameter Combination: you can't keep the destination locked in near-live migration mode"
        end_process 255
    fi

    if [ "$opt_vm_ids" = "all" ]; then
        local all=''
        local data=''
        local cnt=''
        local ids=''

        all=$(get_vm_ids "$QEMU_CONF_CLUSTER/*$EXT_CONF" "$LXC_CONF_CLUSTER/*$EXT_CONF")
        all=$(echo "$all" | tr ',' "\n")
        opt_exclude_vmids=$(echo "$opt_exclude_vmids" | tr ',' "\n")
        for id in $all; do
            cnt=$(echo $opt_exclude_vmids | grep -o $id|wc -w)
            if [ $cnt == 0 ]; then
                vm_ids=$(echo "$vm_ids$id:$opt_prefix_id$id,")
            fi
        done
        vm_ids=$(echo "$vm_ids" | tr ',' "\n")
    else
        if [ ! -z $opt_prefix_id ]; then
            ids=$(echo "$opt_vm_ids" | tr ',' "\n")
            for id in $ids; do
                vm_ids=$(echo "$vm_ids$id:$opt_prefix_id$id,")
            done
            vm_ids=$(echo "$vm_ids" | tr ',' "\n")
        else
            vm_ids=$(echo "$opt_vm_ids" | tr ',' "\n")
        fi
    fi
    log debug "vm_ids: $vm_ids"
}

human_readable() {
    b=${1:-0}; d=''; s=0; S=(Bytes {K,M,G,T,P,E,Z,Y}iB)
    while ((b > 1024)); do
        d="$(printf ".%02d" $((b % 1024 * 100 / 1024)))"
        b=$((b / 1024))
        let s++
    done
    echo "$b$d ${S[$s]}"
}

function map_vmids_to_host(){
    for node in $(/usr/bin/pvecm nodes | tail +5 | tr -s ' ' | cut -d' ' -f 4)
    do
        for vm in $(ssh root@$node qm list|tail +2|tr -s ' '|cut -f 2 -d' ')
        do
            pvnode[$vm]=$node
        done
    done
}

function map_vmids_to_dsthost(){
    for node in $(ssh $1 /usr/bin/pvecm nodes | tail +5 | tr -s ' ' | cut -d' ' -f 4)
    do
        for vm in $(ssh root@$node qm list|tail +2|tr -s ' '|cut -f 2 -d' ')
        do
            dstpvnode[$vm]=$node
        done
    done
}

function exist_file(){
    local file=''
    for file in $1; do
        [ -e "$file" ] && return 0 || return 1
        break
    done
}

function lookupcephpool() {
    pvehost=$1
    pvepoolname=$2
    res=$(ssh $pvehost cat /etc/pve/storage.cfg | sed -n "/rbd: $pvepoolname/,/^$/p" | grep -E "\s+pool\s" | cut -d " " -f 2)
    echo $res
}

function lookupdatapool() {
    pvehost=$1
    pvepoolname=$2
    res=$(ssh $pvehost cat /etc/pve/storage.cfg | sed -n "/rbd: $pvepoolname/,/^$/p" | grep -E "\s+data-pool\s" | cut -d " " -f 2)
    echo $res
}

function get_vm_ids(){
    local data=''
    local conf=''

    while [ $# -gt 0 ]; do
        for conf in $1; do
            [ ! -e "$conf" ] && break

            conf=$(basename "$conf")
            [ "$data" != '' ] && data="$data,"
            data="$data${conf%.*}"
        done
        shift
    done

    echo "$data"
}

function get_disks_from_config(){
    local disks;
    local file_config=$1

    #disks available for vm/ct
    #exclude no backup
    #read current config
    disks=$(while read -r line; do
                [[ "$line" == "" ]] && break
                echo "$line" 
            done < "$file_config" | \
            grep -P '^(?:((?:efidisk|virtio|ide|scsi|sata|mp)\d+)|rootfs): ' | \
            grep -v -P 'cdrom|none' | \
            grep -v -P 'backup=0' | \
            awk '{ split($0,a,","); split(a[1],b," "); print b[2]}')

    echo "$disks"
}

function log(){
    local level=$1
    shift 1
    local message=$*
    local syslog_msg=''

    case $level in
        debug) 
            if [ $opt_debug -eq 1 ]; then
                echo -e "$(date "+%F %T") DEBUG: $message";
                echo -e "$(date "+%F %T") DEBUG: $message" >> "$LOG_FILE";
            fi    
            ;;

        info) 
            echo -e "$message"; 
            echo -e "$message" | sed -e 's/\x1b\[[0-9;]*m//g' >> "$LOG_FILE";
            syslog_msg=$(echo -e "$message" | sed -e ${restripansicolor})
            [ $opt_syslog -eq 1 ] && logger -t "$PROGNAME" "$syslog_msg"
            ;;

        warn)
            echo -n "$(echoyellow 'WARNING: ')"
            echowhite "$message" 1>&2
            echo -e "$message" | sed -e ${restripansicolor} >> "$LOG_FILE";            
            syslog_msg=$(echo -e "$message" | sed -e ${restripansicolor})
            [ $opt_syslog -eq 1 ] && logger -t "$PROGNAME" -p daemon.warn "$syslog_msg"
            ;;
        
        error)
            echo -n "$(echored 'ERROR: ')"
            echowhite "$message" 1>&2
            echo -e "$message" | sed -e ${restripansicolor} >> "$LOG_FILE";            
            syslog_msg=$(echo -e "$message" | sed -e ${restripansicolor})
            [ $opt_syslog -eq 1 ] && logger -t "$PROGNAME" -p daemon.err "$syslog_msg"
            ;;

        *)  
            echo "$message" 1>&2
            echo -e "$message" | sed -e ${restripansicolor} >> "$LOG_FILE";            
            syslog_msg=$(echo -e "$message" | sed -e ${restripansicolor})
            [ $opt_syslog -eq 1 ] && logger -t "$PROGNAME" "$syslog_msg"
            ;;
    esac
}

function map_source_to_destination_vmid(){
    for vm_id_pair in $vm_ids; do
        svid=$(echo "$vm_id_pair" | sed -r -e "s/^([0-9]+):([0-9]+)$/\1\n/")
        dvid=$(echo "$vm_id_pair" | sed -r -e "s/^([0-9]+):([0-9]+)$/\2\n/")
        svmids="$svmids"$'\n'"$svid"
        dvmids[$svid]=$dvid
    done
}

function mirror() {
    local -i rc=0;
    parse_opts "$@"

    local timestamp; timestamp=$(date +%Y%m%d%H%M%S)
    local xmittype
    local humantime
    local vmname
    local -i xmitrc
    local -i ssrc
    local -i freezerc
    local -i unfreezerc
    local -i startdisk
    local -i enddisk
    local -i startjob
    local -i endjob
    local -i vmcount=0
    local -i diskcount=0
    local -i startdowntime
    local -i enddowntime
    local -i ga_ping

    local disp_perf_freeze_failed
    local disp_perf_ss_failed
    local disp_perf_full_failed
    local disp_perf_diff_failed

    log info "ACTION: $(echowhite Onlinemirror)"
    log info "Start mirror $(date "+%F %T")"
    startjob=$(date +%s)

    get_ceph_version
    log info "Local Ceph Version: $scephversion, Remote Ceph Version: $dcephversion"

    #create pid file
    local pid_file="/var/run/$PROGNAME.pid"
    if [[ -e "$pid_file" ]]; then
        local pid; pid=$(cat "${pid_file}")
        if ps -p "$pid" > /dev/null 2>&1; then
          log error "Process already running with pid ${pid}"
          end_process 1
        fi
    fi
    if ! echo $$ > "$pid_file"; then
        log error "Could not create PID file $pid_file"
        end_process 1
    fi

    scluster=$(grep cluster_name /etc/pve/corosync.conf | cut -d " " -f 4)
    dcluster=$(ssh "$opt_destination" grep cluster_name /etc/pve/corosync.conf | cut -d " " -f 4)

    if [ $opt_migrate -eq 1 ] && [ $opt_noconfirm -eq 0 ]; then
        echo "VM(s) $opt_vm_ids will subsequently be shutdown on [$scluster] and started on [$dcluster]"
        read -p "Do you want to proceed? (yes/no) " yn
        case $yn in 
            yes ) echo ok, we will proceed;;
            no ) echo exiting...;
                exit;;
            * ) echo invalid response;
                exit 1;;
        esac
    fi



    map_source_to_destination_vmid
    map_vmids_to_host
    map_vmids_to_dsthost "$opt_destination"

    if [ "$(check_pool_exist "$opt_pool")" -eq 0 ]; then
        log error "Preflight check: Destination RBD-Pool $opt_pool does not exist."
        end_process 255
    fi

    for vm_id in $svmids; do
        file_config="$PVE_NODES/${pvnode[$vm_id]}/$QEMU/$vm_id.conf"
        if ! exist_file "$file_config"; then
            log error "VM $vm_id - Preflight check: VM $vm_id does not exist on source cluster [$scluster] - skipping to next VM."
            continue
        fi
        ga_ping=$(gaping "$vm_id")
        log debug "ga_ping: $ga_ping"
        if [ "$ga_ping" -eq 255 ] ; then    #vm running but no qemu-guest-agent answering
            log error "VM $vm_id - Preflight check: VM $vm_id on source cluster [$scluster] has no qemu-guest-agent running - skipping to next VM."
            continue
        fi
        (( vmcount++ ))
        local disk=''
        dvmid=${dvmids[$vm_id]}
        vmname=$(cat $PVE_NODES/"${pvnode[$vm_id]}"/$QEMU/"$vm_id".conf | sed -e ''$restripsnapshots'' | grep "name\:" | cut -d' ' -f 2)
        log info "VM $vm_id - Starting mirror for $(echowhite "$vmname")"
        srcvmgenid=$(cat $PVE_NODES/"${pvnode[$vm_id]}"/$QEMU/"$vm_id".conf | sed -e ''$restripsnapshots'' | grep vmgenid | sed -r -e 's/^vmgenid:\s(.*)/\1/')
        dstvmgenid=$(ssh "$opt_destination" cat $PVE_NODES/"${dstpvnode[$dvmid]}"/$QEMU/"$dvmid".conf 2>/dev/null | grep vmgenid | sed -e ''$restripsnapshots'' | sed -r -e 's/^vmgenid:\s(.*)/\1/')
        log info "VM $vm_id - Checking for VM $dvmid on destination cluster $opt_destination $QEMU_CONF_CLUSTER"
        log debug "DVMID:$dvmid srcvmgenid:$srcvmgenid dstvmgenid:$dstvmgenid"
        conf_on_destination=$(ssh "$opt_destination" "ls -d $QEMU_CONF_CLUSTER/$dvmid$EXT_CONF 2>/dev/null")
        [[ "$conf_on_destination" =~ $redstconf ]]
        host_on_destination=${BASH_REMATCH[1]}

        if [ $host_on_destination ]; then
            dststatus=$(ssh root@${dstpvnode[$dvmid]} qm status $dvmid|cut -d' ' -f 2)
            if [ $dststatus == "running" ]; then
                log error "VM is running on Destination Cluster [$dcluster]. bailing out"
                end_process 255
            fi
        fi
        #Check if source VM is running
        srcstatus=$(ssh root@${pvnode[$vm_id]} qm status $vm_id|cut -d' ' -f 2)
        if [ $srcstatus == "running" ] && [ $opt_online -eq 0 ]; then
            log error "Source VM is running .. exiting"
            end_process 1
        fi

        if [ ! "$host_on_destination" ] || [ $opt_overwrite -eq 1 ]; then
            if [ "$host_on_destination" ] && [ $srcvmgenid != $dstvmgenid ]; then
                log error "Source VM genid ($srcvmgenid) doesn't match destination VM genid ($dstvmgenid). This should not happen. Bailing out.."
                end_process 255
            fi
            log info "VM $vm_id - Transmitting Config for VM $vm_id to destination $opt_destination VMID $dvmid"
            rewriteconfig $PVE_NODES/"${pvnode[$vm_id]}"/$QEMU/"$vm_id".conf $opt_destination "$opt_pool" $PVE_NODES/"$opt_destination"/$QEMU/"$dvmid".conf "$dvmid"
            map_vmids_to_dsthost "$opt_destination"
        fi

        #--move so we need to shutdown and remove from ha group?
        if [ $opt_migrate -eq 1 ]; then
            log info "VM $vm_id - Migration requested, shutting down VM on ${pvnode[$vm_id]}"
            if [ "$(get_ha_status "$vm_id")" == "started" ]; then
                log info "VM $vm_id - remove from HA"
                do_run "ha-manager remove $vm_id"
            fi
            do_run "ssh root@${pvnode[$vm_id]} qm shutdown $vm_id >/dev/null"
            startdowntime=$(date +%s)
        fi

        #Lock on source + destination
        if [ $opt_lock -eq 1 ]; then
            do_run "ssh root@""${pvnode[$vm_id]}"" qm set ""$vm_id"" --lock backup" >/dev/null
            log info "VM $vm_id - locked $vm_id [rc:$?] on source"
            do_run "ssh root@""${dstpvnode[$dvmid]}"" qm set ""$dvmid"" --lock backup" >/dev/null
            log info "VM $dvmid - locked $dvmid [rc:$?] on destination"
        fi
        #Freeze fs only if no migration running and qemu-guest-agent okay.
        if [ $opt_migrate -eq 0 ] && [ $ga_ping -eq 0 ]; then
            vm_freeze "$vm_id" "${pvnode[$vm_id]}" >/dev/null
            freezerc=$?
            if [ $freezerc -gt 0 ]; then
                log warn "VM $vm_id - QEMU-Guest could not fsfreeze on guest."
                (( perf_freeze_failed++ ))
            else
                (( perf_freeze_ok++ ))
            fi
        fi
        for disk in $(get_disks_from_config "$file_config"); do
            src_image_spec=$(get_image_spec "$disk")
            create_snapshot "$src_image_spec@$opt_snapshot_prefix$timestamp" 2>/dev/null
            ssrc=$?
            if [ $ssrc -gt 0 ]; then
                log warn "VM $vm_id - rbd snap failed."
                (( perf_ss_failed++ ))
            else
                (( perf_ss_ok++ ))
            fi
        done
        if [ $opt_migrate -eq 0 ]; then
            vm_unfreeze "$vm_id" "${pvnode[$vm_id]}" >/dev/null
            unfreezerc=$?
            if [ $unfreezerc -gt 0 ]; then
                log error "VM $vm_id - QEMU-Guest could not fsunfreeze on guest."
            fi
            if [ ! $opt_keepslock -eq 1 ]; then
                do_run "ssh root@${pvnode[$vm_id]} qm unlock $vm_id" >/dev/null
                log info "VM $vm_id - unlocked source VM $vm_id [rc:$?]"
            fi
        fi
        for disk in $(get_disks_from_config "$file_config"); do
            (( diskcount++ ))
            log debug "VMID: $vm_id Disk: $disk DESTVMID: $dvmid"
            src_image_spec=$(get_image_spec "$disk")
            [ -z "$src_image_spec" ] && continue
            dst_image_spec=$(echo $src_image_spec | sed -r -e "s/(.*\/[a-zA-Z0-9]+\-)([0-9]+)(\-[a-zA-Z0-9]+\-[0-9]+)/\1$dvmid\3/")
            [ -z "$dst_image_spec" ] && continue
            [[ $disk =~ $recephimg ]]
	        src_image_pool_pve=${BASH_REMATCH[1]}
            src_image_pool=$(lookupcephpool "localhost" ${BASH_REMATCH[1]})
            src_image_name=${BASH_REMATCH[2]}
            [[ $dst_image_spec =~ ^.*\/(.*)$ ]]
            dst_image_name=${BASH_REMATCH[1]}-$src_image_pool_pve
            dst_image_pool=$(lookupcephpool $opt_destination $opt_pool)
            dst_data_pool=$(lookupdatapool $opt_destination $opt_pool)
            if [ -n "$dst_data_pool" ]; then
                dst_data_opt="--data-pool $dst_data_pool"
            fi
            snapshot_name="@$opt_snapshot_prefix$timestamp"
            localsnapcount=$(rbd ls -l $src_image_pool | grep $src_image_name@$opt_snapshot_prefix | cut -d ' ' -f 1|wc -l)
            if [ $localsnapcount -ge 2 ]; then
            # we have at least 2 local snapshots, to we can make an incremental copy
                currentlocal=$(rbd ls -l $src_image_pool | grep $src_image_name@$opt_snapshot_prefix | cut -d ' ' -f 1|tail -n 1)
                localts=$(rbd ls -l $src_image_pool | grep $src_image_name@$opt_snapshot_prefix | cut -d ' ' -f 1 | sed -r -e 's/.*@'$opt_snapshot_prefix'(.*)/\1/')
            fi
            latestremote=$(ssh $opt_destination rbd ls -l $dst_image_pool | grep $dst_image_name@$opt_snapshot_prefix | cut -d ' ' -f 1|tail -n 1)
            if [ "$latestremote" ]; then
                [[ $latestremote =~ ^.*@$opt_snapshot_prefix([0-9]+)$ ]]
                latestremotets=${BASH_REMATCH[1]}
                for ts in $localts; do
                    if [ "$ts" == "$latestremotets" ]; then
                        basets=$ts
                    fi
                done
            fi
            if [ -z "$basets" ]; then
                xmittype='full'
                log debug "No matching Snapshot found on destination - Full Copy $src_image_pool/$src_image_name$snapshot_name to $dst_image_pool/$dst_image_name"
                #snapts=$(echo $currentlocal | sed -r -e 's/.*@mirror-(.*)/\1/')
                snapshotsize=$(rbd du --pretty-format --format json $src_image_pool/$src_image_name|jq '.images[] | select (.snapshot_id == null) | {provisioned_size}.provisioned_size'|tail -1)
                log debug "snapsize: $snapshotsize "
                xmitjob="rbd export --rbd-concurrent-management-ops 8 $src_image_pool/$src_image_name$snapshot_name --no-progress - | tee >({ wc -c; } >/tmp/$PROGNAME.$pid.$dst_image_pool-$dst_image_name.size) | pv -s $snapshotsize -F \"VM $vm_id - F $src_image_pool/$src_image_name$snapshot_name: $PVFORMAT_FULL\" | ssh $opt_destination rbd import --image-format 2 - $dst_image_pool/$dst_image_name $dst_data_opt 2>/dev/null"
                # create initial snapshot on destination
                log debug "xmitjob: $xmitjob"
                startdisk=$(date +%s)
                do_run "$xmitjob"
                enddisk=$(date +%s)
                xmitrc=$?
                if [ ! $xmitrc ]; then
                    log error "Transmitting Image failed"
                    (( perf_full_failed++ ))
                    return 1
                else
                    (( perf_full_ok++ ))
                fi
                cmd="ssh $opt_destination rbd snap create $dst_image_pool/$dst_image_name$snapshot_name"
                do_run "$cmd" 2>/dev/null
                log info "VM $vm_id - created snapshot on $dvmid [rc:$?]"
                perf_bytes_full=$(( perf_bytes_full + $(cat /tmp/"$PROGNAME"."$pid"."$dst_image_pool"-"$dst_image_name".size) ))
            else
                xmittype='incremental'
                log debug "Basecopy + snapshot on destination - let's just transfer the diff"
                log debug "sizer: rbd diff $src_image_pool/$currentlocal --from-snap $opt_snapshot_prefix$basets|gawk --bignum '{ SUM += \$2 } END { print SUM }'"
                snapshotsize=$(rbd diff $src_image_pool/$currentlocal --from-snap $opt_snapshot_prefix$basets|gawk --bignum '{ SUM += $2 } END { print SUM }')
                log debug "snapshotsize: $snapshotsize"
                if [ -z "$snapshotsize" ]; then
                    #disk was not attached, or really nothing has changed..
                    snapshotsize=0
                fi
                xmitjob="rbd export-diff --no-progress --from-snap $opt_snapshot_prefix$basets $src_image_pool/$currentlocal - | tee >({ wc -c; } >/tmp/$PROGNAME.$pid.$dst_image_pool-$dst_image_name.size) | pv -F \"VM $vm_id - I $src_image_pool/$src_image_name$snapshot_name: $PVFORMAT_SNAP\" | ssh $opt_destination rbd import-diff --no-progress - $dst_image_pool/$dst_image_name $dst_data_opt"
                log debug "xmitjob: $xmitjob"
                startdisk=$(date +%s)
                do_run "$xmitjob"
                enddisk=$(date +%s)
                xmitrc=$?
                if [ ! $xmitrc ]; then
                    log error "Transmitting Image failed"
                    (( perf_diff_failed++ ))
                    return 1
                else
                    (( perf_diff_ok++ ))
                fi
                do_housekeeping "localhost" "$src_image_pool" "$src_image_name" "$opt_keep_local" "$vm_id"
                do_housekeeping "$opt_destination" "$dst_image_pool" "$dst_image_name" "$opt_keep_remote" "$vm_id"
                perf_bytes_diff=$(( perf_bytes_diff + $(cat /tmp/"$PROGNAME"."$pid"."$dst_image_pool"-"$dst_image_name".size) ))
            fi
            perf_bytes_total=$(( perf_bytes_total + $(cat /tmp/"$PROGNAME"."$pid"."$dst_image_pool"-"$dst_image_name".size) ))
            rm /tmp/"$PROGNAME"."$pid"."$dst_image_pool"-"$dst_image_name".size
            log info "VM $vm_id - Disk Summary: Took $(( enddisk - startdisk )) Seconds to transfer $(human_readable "$perf_bytes_total" 2) in a $xmittype run"
            if [ -n "$opt_influx_api_url" ]; then
                log info "VM $vm_id - Logging to InfluxDB: $opt_influx_api_url"
                influxlp="$opt_influx_job_metrics,vmname=$vmname,jobname=$opt_influx_jobname,destination=$opt_destination,srcimage=$src_image_name,dstimage=$dst_image_name,xmittype=$xmittype bytescalculated=$snapshotsize""i,bytesonwire=$perf_bytes_total""i,xmitrc=$xmitrc""i,freezerc=$freezerc""i,unfreezerc=$unfreezerc""i,basets=$basets""i"
                log debug "InfluxLP: --->\n $influxlp"
                cmd="curl --request POST \"$opt_influx_api_url/v2/write?org=$opt_influx_api_org&bucket=$opt_influx_bucket&precision=ns\" --header \"Authorization: Token $opt_influx_token\" --header \"Content-Type: text/plain; charset=utf-8\" --header \"Accept: application/json\" --data-binary '$influxlp'"
                do_run "$cmd"
            fi
            unset basets
        done
        if [ $opt_keepdlock -eq 0 ]; then
            ssh root@${dstpvnode[$dvmid]} qm unlock $dvmid
            log info "VM $dvmid - Unlocking destination VM $dvmid"
        fi
        #--migrate so start on destination?
        if [ $opt_migrate -eq 1 ]; then
            log info "VM $dvmid - Starting VM on node ${dstpvnode[$dvmid]} in cluster [$dcluster]"
            do_run "ssh root@""${dstpvnode[$dvmid]}"" qm start "$dvmid >/dev/null
            enddowntime=$(date +%s)
            log info "VM $dvmid - Downtime: $(( enddowntime - startdowntime )) Seconds"
        fi

    done
    endjob=$(date +%s)
    log info "Finnished mirror $(date "+%F %T")"
    humantime=$(date -ud "@$((endjob-startjob))" +'%H hours %M minutes %S seconds')
    log info "Job Summary: Bytes transferred $(human_readable $perf_bytes_total) for $diskcount Disks on $vmcount VMs in $humantime"
    if [ "$perf_freeze_failed" -gt 0 ]; then disp_perf_freeze_failed="$(echored $perf_freeze_failed)"; else disp_perf_freeze_failed="$(echogreen $perf_freeze_failed)"; fi
    if [ "$perf_ss_failed" -gt 0 ]; then disp_perf_ss_failed="$(echored $perf_ss_failed)"; else disp_perf_ss_failed="$(echogreen $perf_ss_failed)"; fi
    if [ "$perf_full_failed" -gt 0 ]; then disp_perf_full_failed="$(echored $perf_full_failed)"; else disp_perf_full_failed="$(echogreen $perf_full_failed)"; fi
    if [ "$perf_diff_failed" -gt 0 ]; then disp_perf_diff_failed="$(echored $perf_diff_failed)"; else disp_perf_diff_failed="$(echogreen $perf_diff_failed)"; fi
    log info "VM Freeze OK/failed.......: $perf_freeze_ok/$disp_perf_freeze_failed"    
    log info "RBD Snapshot OK/failed....: $perf_ss_ok/$disp_perf_ss_failed"
    log info "RBD export-full OK/failed.: $perf_full_ok/$disp_perf_full_failed"
    log info "RBD export-diff OK/failed.: $perf_diff_ok/$disp_perf_diff_failed"
    log info "Full xmitted..............: $(human_readable $perf_bytes_full)"
    log info "Differential Bytes .......: $(human_readable $perf_bytes_diff)"
    if [ -n "$opt_influx_api_url" ]; then
        log info "VM $vm_id - Logging Job summary to InfluxDB: $opt_influx_api_url"
        influxlp="$opt_influx_summary_metrics,jobname=$opt_influx_jobname perf_bytes_diff=$perf_bytes_diff""i,perf_bytes_full=$perf_bytes_full""i,perf_bytes_total=$perf_bytes_total""i,perf_diff_failed=$perf_diff_failed""i,perf_diff_ok=$perf_diff_ok""i,perf_freeze_failed=$perf_freeze_failed""i,perf_freeze_ok=$perf_freeze_ok""i,perf_full_failed=$perf_full_failed""i,perf_full_ok=$perf_full_ok""i,perf_ss_failed=$perf_ss_failed""i,perf_ss_ok=$perf_ss_ok""i,perf_vm_running=$perf_vm_running""i,perf_vm_stopped=$perf_vm_stopped""i"
        log debug "InfluxLP: --->\n $influxlp"
        cmd="curl --request POST \"$opt_influx_api_url/v2/write?org=$opt_influx_api_org&bucket=$opt_influx_bucket&precision=ns\" --header \"Authorization: Token $opt_influx_token\" --header \"Content-Type: text/plain; charset=utf-8\" --header \"Accept: application/json\" --data-binary '$influxlp'"
        do_run "$cmd"
    fi
    (( perf_vm_ok++ ))
    end_process 0
}

function do_housekeeping(){
    local horst=$1
    local rbdpool=$2
    local rbdimage=$3
    local keep=$4
    local vm=$5
    local snap
    local -i keeptime
    local -i ts
    local -i snapepoch
    local -i age

    log info "VM $vm - Housekeeping: $horst $rbdpool/$rbdimage, keeping Snapshots for $keep"
    cmd="ssh $horst rbd ls -l $rbdpool | grep $rbdimage@$opt_snapshot_prefix | cut -d ' ' -f 1|head -n -1"
    snapshots=$($cmd)
    if [ "${keep:(-1)}" == "d" ]; then
        keep=${keep%?}
        keeptime=$(( $keep * 86400 ))
    elif [ "${keep:(-1)}" == "s" ]; then
        keep=${keep%?}
        keeptime=$keep
    fi
    for snap in $snapshots; do
        [[ $snap =~ ^.*@$opt_snapshot_prefix([0-9]+)$ ]]
        ts=${BASH_REMATCH[1]}
        [[ $ts =~ $redateex ]]
        snapepoch=$(date --date "${BASH_REMATCH[1]}/${BASH_REMATCH[2]}/${BASH_REMATCH[3]} ${BASH_REMATCH[4]}:${BASH_REMATCH[5]}:${BASH_REMATCH[6]}" +%s)
        age=$(($(date -u +"%s")-$snapepoch ))
        if [ $age -gt "$keeptime" ]; then
            cmd="ssh $horst rbd snap rm $rbdpool/$snap"
            do_run "$cmd" 2>/dev/null
            log info "VM $vm_id - Removing Snapshot $horst $rbdpool/$snap ($age""s) [rc:$?]"
            if [ $rc -eq 0 ]; then
                (( perf_snaps_removed++ ))
            fi
        fi
    done
}

function gaping() {
    local vmid=$1
    local rc
    cmd="ssh root@${pvnode[$vmid]} qm guest cmd $vmid ping >/dev/null 2>&1"
    eval "$cmd"
    rc=$?
    echo $rc
}

function create_snapshot(){
    local snap="$1"
    log info "VM $vm_id - Creating snapshot $snap"
    do_run "rbd snap create $snap"
    rc=$?
    log debug "create_snapshot() return $rc"
    return $rc
}

function vm_freeze() {
    local fvm=$1;
    local fhost=$2;
    status=$(ssh root@"$fhost" qm status "$fvm"|cut -d' ' -f 2)
    if ! [[ "$status" == "running" ]]; then
        log info "VM $fvm - Not running, skipping fsfreeze-freeze"
        (( perf_vm_stopped++ ))
        return
    else 
        (( perf_vm_running++ ))
    fi
    local cmd="ssh root@$fhost /usr/sbin/qm guest cmd $fvm fsfreeze-freeze"
    log info "VM $fvm - Issuing fsfreeze-freeze to $fvm on $fhost"
    do_run "$cmd"
    rc=$?
    log debug "vm_freeze() return $rc"
    return $rc
}

function vm_unfreeze() {
    local fvm=$1;
    local fhost=$2;
    status=$(ssh root@"$fhost" qm status "$fvm"|cut -d' ' -f 2)
    if ! [[ "$status" == "running" ]]; then
        log info "VM $fvm - Not running, skipping fsfreeze-thaw"
        return
    fi
    local cmd="ssh root@$fhost /usr/sbin/qm guest cmd $fvm fsfreeze-thaw"
    log info "VM $fvm - Issuing fsfreeze-thaw to $fvm on $fhost"
    do_run "$cmd"
    rc=$?
    log debug "vm_unfreeze() return $rc"
    return $rc
}

function rewriteconfig(){
    local oldconfig=$1
    local dst=$2
    local newpool=$3
    local newconfig=$4
    local newvmid=$5
    local sedcmd
    if [ ! -z "$opt_rewrite" ]; then
        sedcmd='sed -r -e '$opt_rewrite
    else
        sedcmd='sed -e /^$/,$d'
    fi
    cat "$oldconfig" | sed -r -e "s/^(efidisk|virtio|ide|scsi|sata|mp)([0-9]+):\s([a-zA-Z0-9]+):(.*)-([0-9]+)-disk-([0-9]+).*,(.*)$/\1\2: $newpool:\4-$newvmid-disk-\6-\3,\7/g" | $sedcmd | sed -e '/^$/,$d' | sed -e '/ide[0-9]:.*-cloudinit,media=cdrom.*/d' | grep -v "^parent:\s.*$" | ssh "$dst" "cat - >$newconfig"
}

function checkvmid(){
    local dst=$1
    local vmid=$2
    cmd="ssh $dst ls -l $QEMU_CONF_CLUSTER/$vmid.conf|wc -l"
    rval=$($cmd)
    echo "$rval"
}

function do_run(){
    local cmd=$*;
    local -i rc=0;

    if [ $opt_dry_run -eq 1 ]; then
        echo "$cmd"
        rc=$?
    else
        log debug "$cmd"
        eval "$cmd"
        rc=$?
        [ $rc != 0 ] && log error "$cmd"
        log debug "return $rc ps ${PIPESTATUS[@]}"
    fi

    return $rc
}

function end_process(){
    local -i rc=$1;
    local -i runtime
    local -i bps
    local -i ss_total
    local subject
    if ! [[ -z "$startjob" || -z "$endjob" ]]; then
        runtime=$(expr $endjob - $startjob)
        bps=$(expr $perf_bytes_total/$runtime)
    fi
    ss_total=$(expr $perf_ss_ok + $perf_ss_failed)
    subject="Crossover [VM:$perf_vm_ok/$vmcount SS:$perf_ss_ok/$ss_total]"
    [ $rc != 0 ] && subject="[ERROR] $subject" || subject="[OK] $subject"

    local mail;
    for mail in $(echo "$opt_addr_mail" | tr "," "\n"); do
        do_run "cat '$LOG_FILE' | mail -s '$subject' '$mail'"
    done

    #remove log
    rm "$LOG_FILE"
    exit "$rc";
}

function get_image_spec(){
    local image_spec;
    local disk="$1"
    
    #if krbd enable
    image_spec=$(pvesm path "$disk" | grep '^/dev/rbd/' | sed -e "s/^\/dev\/rbd\///")
    if [ -z "$image_spec" ]; then
        image_spec=$(pvesm path "$disk" | grep '/ceph/' | awk '{ split($0,a,":"); print a[2]}')
    fi

    echo "$image_spec"
}

function get_ha_status() {
    local havmid="$1"
    ha_status=$(ha-manager status| grep vm:"$havmid" | cut -d " " -f 4| sed 's/.$//')
    echo "$ha_status"
}

function check_pool_exist() {
    local poolname="$1"
    local -i exists=255
    pool_status=$(ssh $opt_destination pvesm status|grep rbd|cut -d " " -f 1|grep $poolname)
    if [ "$pool_status" == "$poolname" ]; then
        exists=1
    else
        exists=0
    fi
    echo $exists
}

function get_ceph_version() {
    scephversion=$(ceph -v | cut -d " " -f 3)
    dcephversion=$(ssh $opt_destination ceph -v | cut -d " " -f 3)
}

function main(){    
    [ $# = 0 ] && usage;

    #command
    case "$1" in
        version) echo "$VERSION";;
        help) usage "$@";;
        mirror) mirror "$@";; 
        *) usage;;
    esac

    exit 0;
}

main "$@"