Add support for ceph erasure-coded pools, fix a bug, when a pool was called pool

README.md

@@ -90,7 +90,8 @@ It'll work according this scheme:
 
 ## Installation of prerequisites
 
-```apt install git pv gawk jq
+```
+apt install git pv gawk jq curl
 
 ## Install the Script somewhere, eg to /opt
 
@@ -222,13 +223,13 @@ Differential Bytes .......: 4.37 MiB
 
 ## Things to check
 
-From Proxmox VE Hosts you want to backup you need to be able to ssh passwordless to all other Cluster hosts, that may hold VM's or Containers. This goes for the source and for the destination Cluster.
+From Proxmox VE Hosts you want to backup you need to be able to ssh passwordless to all other Cluster hosts, that may hold VM's or Containers. This goes for the source and for the destination Cluster. Doublecheck this.
 
 This is required for using the free/unfreeze and the lock/unlock function, which has to be called locally from that Host the guest is currently running on. Usually this works out of the box for the source cluster, but you may want to make sure that you can "ssh root@pvehost1...n" from every host to every other host in the cluster.
 
-For the Destination Cluster you need to copy your ssh-key to the first host in the cluster, and login once to every node
+For the Destination Cluster you need to copy your ssh-key to the first host in the cluster, and login once to every node in your cluster.
-in your cluster.
 
+Currently preflight checks don't include the check for enough resources in the destination cluster. Check beforehand that you don't exceed the maximum safe size of ceph in the destination cluster.
 
 ## Some words about Snapshot consistency and what qemu-guest-agent can do for you
 
@@ -371,3 +372,4 @@ Ceph Documentation:
 
 Proxmox Wiki:
 https://pve.proxmox.com/wiki/
+

crossover

@@ -1,5 +1,8 @@
 #!/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
@@ -13,14 +16,11 @@ declare opt_influx_jobname=''
 declare opt_influx_job_metrics='crossover_xmit'
 declare opt_influx_summary_metrics='crossover_jobs'
 
-# Cross Pool Migration and incremental replication Tool for Proxmox VMs using Ceph.
-# Author: Bastian Mäuser <bma@netz.org>        
-
 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'
@@ -28,15 +28,21 @@ 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
@@ -56,10 +62,12 @@ 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
@@ -71,6 +79,7 @@ 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'
 
@@ -113,7 +122,7 @@ Options:
     --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
+    --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
@@ -130,6 +139,7 @@ Switches:
     --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>
@@ -144,7 +154,7 @@ function parse_opts(){
     local args
     args=$(getopt \
            --options '' \
-           --longoptions=vmid:,prefixid:,excludevmids:,destination:,pool:,keeplocal:,keepremote:,rewrite:,influxurl:,influxorg:,influxtoken:,influxbucket:,jobname:,mail:,online,migrate,nolock,keep-slock,keep-dlock,overwrite,dry-run,debug,syslog \
+           --longoptions=vmid:,prefixid:,excludevmids:,destination:,pool:,keeplocal:,keepremote:,rewrite:,influxurl:,influxorg:,influxtoken:,influxbucket:,jobname:,mail:,online,migrate,nolock,keep-slock,keep-dlock,overwrite,dry-run,noconfirm,debug,syslog \
            --name "$PROGNAME" \
            -- "$@") \
            || end_process 128
@@ -170,6 +180,7 @@ function parse_opts(){
         --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;;
@@ -287,7 +298,14 @@ function exist_file(){
 function lookupcephpool() {
     pvehost=$1
     pvepoolname=$2
-    res=$(ssh $pvehost cat /etc/pve/storage.cfg | sed -n "/rbd: $pvepoolname/,/^$/p" | grep pool | cut -d " " -f 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
 }
 
@@ -309,20 +327,6 @@ function get_vm_ids(){
     echo "$data"
 }
 
-function get_config_file(){
-    local file_config=''
-
-    if exist_file "$QEMU_CONF_CLUSTER/$vm_id$EXT_CONF"; then
-        file_config=$(ls $QEMU_CONF_CLUSTER/$vm_id$EXT_CONF)
-
-    else
-        log error "VM $vm_id - Unknown technology or VMID not found: $QEMU_CONF_CLUSTER/$vm_id$EXT_CONF"
-        end_process 128
-    fi 
-
-    echo "$file_config"
-}
-
 function get_disks_from_config(){
     local disks;
     local file_config=$1
@@ -334,7 +338,7 @@ function get_disks_from_config(){
                 [[ "$line" == "" ]] && break
                 echo "$line" 
             done < "$file_config" | \
-            grep -P '^(?:((?:virtio|ide|scsi|sata|mp)\d+)|rootfs): ' | \
+            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]}')
@@ -415,6 +419,9 @@ function mirror() {
     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
@@ -425,6 +432,9 @@ function mirror() {
     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
@@ -439,6 +449,23 @@ function mirror() {
         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"
@@ -449,9 +476,18 @@ function mirror() {
     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 file_config; file_config=$(get_config_file)
-        [ -z "$file_config" ] && continue
         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)
@@ -467,7 +503,7 @@ function mirror() {
         if [ $host_on_destination ]; then
             dststatus=$(ssh root@${dstpvnode[$dvmid]} qm status $dvmid|cut -d' ' -f 2)
             if [ $dststatus == "running" ]; then
-                log error "VM is running on Destination. bailing out"
+                log error "VM is running on Destination Cluster [$dcluster]. bailing out"
                 end_process 255
             fi
         fi
@@ -483,7 +519,7 @@ function mirror() {
                 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 to destination $opt_destination VMID $dvmid"
+            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
@@ -496,6 +532,7 @@ function mirror() {
                 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
@@ -505,8 +542,8 @@ function mirror() {
             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
+        #Freeze fs only if no migration running and qemu-guest-agent okay.
-        if [ $opt_migrate -eq 0 ]; then
+        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
@@ -552,6 +589,10 @@ function mirror() {
             [[ $dst_image_spec =~ ^.*\/(.*)$ ]]
             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 [ $localsnapcount -ge 2 ]; then
@@ -575,7 +616,7 @@ function mirror() {
                 #snapts=$(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  2>/dev/null"
+                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)
@@ -603,7 +644,7 @@ function mirror() {
                     #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"
+                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"
@@ -638,8 +679,10 @@ function mirror() {
         fi
         #--migrate so start on destination?
         if [ $opt_migrate -eq 1 ]; then
-            log info "VM $dvmid - Starting VM on ${pvnode[$vm_id]}"
+            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
@@ -707,6 +750,15 @@ function do_housekeeping(){
     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 $snap"
@@ -722,8 +774,8 @@ function vm_freeze() {
     status=$(ssh root@"$fhost" qm status "$fvm"|cut -d' ' -f 2)
     if ! [[ "$status" == "running" ]]; then
         log info "VM $fvm - Not running, skipping fsfreeze-freeze"
-        return
         (( perf_vm_stopped++ ))
+        return
     else 
         (( perf_vm_running++ ))
     fi
@@ -763,7 +815,7 @@ function rewriteconfig(){
     else
         sedcmd='sed -e /^$/,$d'
     fi
-    cat "$oldconfig" | sed -r -e "s/^(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"
+    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(){
@@ -798,7 +850,7 @@ function end_process(){
     local -i bps
     local -i ss_total
     local subject
-    if ! [[ -z "$startjob" && -z "$endjob" ]]; then
+    if ! [[ -z "$startjob" || -z "$endjob" ]]; then
         runtime=$(expr $endjob - $startjob)
         bps=$(expr $perf_bytes_total/$runtime)
     fi
@@ -847,6 +899,11 @@ function check_pool_exist() {
     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;