iankelling.org Git - distro-setup/blob - btrbk-run

   1 #!/bin/bash
   2 # Copyright (C) 2016 Ian Kelling
   3
   4 # Licensed under the Apache License, Version 2.0 (the "License");
   5 # you may not use this file except in compliance with the License.
   6 # You may obtain a copy of the License at
   7
   8 #     http://www.apache.org/licenses/LICENSE-2.0
   9
  10 # Unless required by applicable law or agreed to in writing, software
  11 # distributed under the License is distributed on an "AS IS" BASIS,
  12 # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
  13 # See the License for the specific language governing permissions and
  14 # limitations under the License.
  15
  16
  17 # todo: if we cancel in the middle of a btrfs send, then run again immediately, the received subvolume doesn't get a Received UUID: field, and we won't mount it. Need to figure out a solution that will fix this.
  18
  19 set -eE -o pipefail
  20 trap 'echo "$0:$LINENO:error: \"$BASH_COMMAND\" returned $?" >&2' ERR
  21
  22 [[ $EUID == 0 ]] || exec sudo -E "$BASH_SOURCE" "$@"
  23
  24 usage() {
  25   cat <<'EOF'
  26 btrbk-run [OPTIONS]
  27 usually -t TARGET_HOST or -s SOURCE_HOST
  28
  29 Note, at source location, intentionally not executable, run and read
  30 install-my-scripts.
  31
  32 EOF
  33   echo "top of script file:"
  34   sed -n '1,/^[# ]*end command line/{p;b};q' "$0"
  35   exit $1
  36 }
  37
  38 # latest $MAIL_HOST
  39 if [[ -e /b/bash_unpublished/source-semi-priv ]]; then
  40   source /b/bash_unpublished/source-semi-priv
  41 fi
  42
  43 script_dir=$(dirname $(readlink -f "$BASH_SOURCE"))
  44
  45 # note q is owned by root:1000
  46 # note p/m is owned 1000:1000 and chmod 700
  47
  48
  49 mountpoints=()
  50
  51 rsync_mountpoint=/q
  52
  53 # default options
  54 conf_only=false
  55 dry_run=false # mostly for testing
  56 rate_limit=no
  57 verbose=true; verbose_arg=-v
  58 progress_arg="--progress"
  59
  60 default_args_file=/etc/btrbk-run.conf
  61 if [[ -s $default_args_file ]]; then
  62   set -- $(< $default_args_file) "$@"
  63 fi
  64
  65 temp=$(getopt -l help cl:m:nps:t:vh "$@") || usage 1
  66 eval set -- "$temp"
  67 while true; do
  68   case $1 in
  69     # only creates the config file, does not run btrbk
  70     -c) conf_only=true; shift ;;
  71     # bytes per second, suffix k m g
  72     -l) rate_limit=$2; shift 2 ;;
  73     # Comma separated mountpoints to backup. This has defaults set below.
  74     -m) IFS=, mountpoints=($2); unset IFS; shift 2 ;;
  75     -n) dry_run=true; dry_run_arg=-n; shift ;;
  76     -p) progress_arg="--progress"; shift ;;
  77     -q) verbose=false; verbose_arg=; progress_arg=; shift ;;
  78     # source host to receive a backup from
  79     -s) source=$2; shift 2 ;;
  80     # target hosts to send to. empty is valid for just doing local
  81     # snapshot. we have default hosts we will populate.
  82     -t) IFS=, targets=($2); unset IFS; shift 2 ;;
  83     -v) verbose=true; verbose_arg=-v; shift ;;
  84     -h|--help) usage ;;
  85     --) shift; break ;;
  86     *) echo "$0: Internal error!" ; exit 1 ;;
  87   esac
  88 done
  89
  90 # usefull commands are resume and archive
  91 cmd_arg=${1:-run}
  92
  93 if [[ -s $default_args_file ]]; then
  94   echo "$0: warning: default btrbk-run options set in $default_args_file (sleeping 5 seconds):"
  95   cat $default_args_file
  96   sleep 5
  97 fi
  98
  99 if [[ -v targets && $source ]]; then
 100   echo "$0: error: -t and -s are mutually exclusive" >&2
 101   exit 1
 102 fi
 103
 104 echo -e "$0: options: conf_only=$conf_only\ndry_run=$dry_run\nrate_limit=$rate_limit\nverbose=$verbose\ncmd_arg=$cmd_arg"
 105
 106 # set default targets
 107 if [[ ! -v targets && ! $source ]]; then
 108   case $HOSTNAME in
 109     x2|kw)
 110       if [[ $HOSTNAME == "$MAIL_HOST" ]]; then
 111         targets=($HOME_DOMAIN)
 112       fi
 113       ;;
 114     tp)
 115       targets=(frodo)
 116       if [[ $HOSTNAME == "$MAIL_HOST" ]]; then
 117         if timeout -s 9 10 ssh x2 :; then
 118           targets+=(x2)
 119         fi
 120       fi
 121       ;;
 122     frodo)
 123       targets=()
 124       ;;
 125     *)
 126       echo "$0: error: no default targets for this host, use -t"
 127       exit 1
 128       ;;
 129   esac
 130 fi
 131
 132 if [[ -v targets ]]; then
 133   echo "targets: ${targets[*]}"
 134 fi
 135
 136 if [[ $source ]]; then
 137   echo "source: $source"
 138 fi
 139
 140
 141 if [[ $mountpoints ]]; then
 142   for mp in ${mountpoints[@]}; do # default mountpoints to sync
 143     if [[ -e /nocow/btrfs-stale/$mp ]]; then
 144       echo "$0: error: $mp is stale, mount-latest-subvol first"
 145       exit 1
 146     fi
 147   done
 148 else
 149   # set default mountpoints
 150   case $HOSTNAME in
 151     # no remote backups atm. note, if we do enable this, configuration below will need some changes.
 152     #  frodo)
 153     #    prospective_mps=(/i)
 154     #    ;;
 155     *)
 156       prospective_mps=(/a /q)
 157       if [[ $HOSTNAME == "$MAIL_HOST" ]]; then
 158         prospective_mps+=(/o)
 159       fi
 160       ;;
 161   esac
 162   for mp in ${prospective_mps[@]}; do # default mountpoints to sync
 163     if [[ -e /nocow/btrfs-stale/$mp ]]; then
 164       echo "$0: warning: $mp stale, not adding to default mountpoints"
 165       continue
 166     fi
 167     if awk '{print $2}' /etc/fstab | grep -xF $mp &>/dev/null; then
 168       mountpoints+=($mp)
 169     fi
 170   done
 171 fi
 172
 173 echo "mountpoints: ${mountpoints[*]}"
 174
 175 ##### end command line parsing ########
 176
 177 rsync-dirs() {
 178   local host=$1
 179   local path=$2
 180   m rsync $dry_run_arg -ahi --relative --delete "$path" "root@$host:/"
 181 }
 182
 183
 184 m() { printf "%s: %s\n" "${0##*/}" "$*";  "$@"; }
 185
 186
 187 if ! which btrbk &>/dev/null; then
 188   echo "$0: error: no btrbk binary found"
 189   exit 1
 190 fi
 191 # if our mountpoints are from stale snapshots,
 192 # it doesn't make sense to do a backup.
 193 check-subvol-stale ${mountpoints[@]} || exit 1
 194
 195
 196 cat >/etc/btrbk.conf <<EOF
 197 ssh_identity /root/.ssh/home
 198 # Just a guess that local7 is a good facility to pick.
 199 # It's a bit odd that the transaction log has to be logged to
 200 # a file or syslog, while other output is sent to std out.
 201 # The man does not mention a way for them to be together, but
 202 # I dunno if setting a log level like warn might also output
 203 # transaction info.
 204 transaction_syslog local7
 205
 206 # 20%ish speedup[]
 207 stream_buffer 512m
 208
 209 # so we only run one at a time
 210 lockfile                   /var/lock/btrbk.lock
 211
 212 # default format of short does not accomidate hourly preservation setting
 213 timestamp_format long-iso
 214
 215 # only make a snapshot if things have changed
 216 snapshot_create onchange
 217 # I could make this different from target_preserve,
 218 # if one disk had less space.
 219 # for now, keeping them equal.
 220 snapshot_preserve 36h 14d 8w 24m
 221 snapshot_preserve_min 4h
 222 snapshot_dir btrbk
 223
 224 # so, total backups = ~89
 225 target_preserve 36h 14d 8w 24m
 226 target_preserve_min 4h
 227
 228 # if something fails and it's not obvious, try doing
 229 # btrbk -l debug -v dryrun
 230
 231 rate_limit $rate_limit
 232 EOF
 233
 234
 235
 236 for tg in ${targets[@]:-$HOSTNAME}; do
 237   # for an initial run, btrbk requires the dir to exist.
 238   ssh root@$tg mkdir -p /mnt/root/btrbk
 239 done
 240
 241
 242
 243 vol=/mnt/root
 244 for m in ${mountpoints[@]}; do
 245   sub=${m##*/}
 246   if [[ $source ]]; then
 247     cat >>/etc/btrbk.conf <<EOF
 248 volume ssh://$source$vol
 249 subvolume $sub
 250 target send-receive $vol/btrbk
 251 EOF
 252   else
 253     cat >>/etc/btrbk.conf <<EOF
 254 volume $vol
 255 subvolume $sub
 256 EOF
 257     for tg in ${targets[@]}; do
 258       cat >>/etc/btrbk.conf <<EOF
 259 target send-receive ssh://$tg$vol/btrbk
 260 EOF
 261     done
 262   fi
 263 done
 264
 265
 266 # todo: umount first to ensure we don't have any errors
 267 # todo: do some kill fuser stuff to make umount more reliable
 268
 269
 270 if $conf_only; then
 271   exit
 272 fi
 273
 274
 275
 276 if $dry_run; then
 277   m btrbk -v -n $cmd_arg
 278   exit 0
 279 elif [[ $cmd_arg == archive ]]; then
 280   if [[ $source ]]; then
 281     m btrbk $verbose_arg $progress_arg $cmd_arg ssh://$source$vol $vol
 282   else
 283     for tg in ${targets[@]}; do
 284       m btrbk $verbose_arg $progress_arg $cmd_arg $vol ssh://$tg$vol
 285     done
 286   fi
 287   exit 0
 288 fi
 289 # -q and just using the syslog option seemed nice,
 290 # but it doesn't show when a send has a parent and when it doesn't.
 291 m btrbk $verbose_arg $progress_arg $cmd_arg
 292
 293 # if we have it, sync to systems which don't
 294 if mountpoint $rsync_mountpoint >/dev/null; then
 295   for tg in ${targets[@]}; do
 296     case $tg in
 297       li|lk)
 298         for x in /p/c/machine_specific/*.hosts; do
 299           if grep -qxF $tg $x; then
 300             dir=${x%.hosts}
 301             rsync-dirs $tg $dir
 302           fi
 303         done
 304         ;;
 305     esac
 306   done
 307 fi
 308
 309 /a/bin/distro-setup/install-my-scripts
 310 if [[ $source ]]; then
 311   m mount-latest-subvol
 312 else
 313   m /a/exe/mount-latest-remote ${targets[@]}
 314 fi
 315
 316 # todo: move variable data we don't care about backing up
 317 # to /nocow and symlink it.
 318
 319
 320 # background on btrbk timezones. with short/long, timestamps use local time.
 321 # for long, if your local time moves backwards, by moving timezones or
 322 # for an hour when daylight savings changes it, you will temporarily get
 323 # a more aggressive retention policy for the overlapping period, and
 324 # vice versa for the opposite timezone move. The alternative is using
 325 # long-iso, which puts timezone info into the timestamp, which means
 326 # that instead of shifting time, you shift the start of day/week/month
 327 # which is used for retention to your new local time, which means for
 328 # example, if you moved forward by 8 hours, the daily/weekly/monthly
 329 # retention will be 8 hours more aggressive since midnight is at a new
 330 # time, unless you fake the timzeone using the TZ env variable.
 331 # However, in the short term, there will be no inconsistencies.
 332 # I don't see any problem with shifting when the day starts for
 333 # retention, so I'm using long-iso.
 334
 335 # note to create a long-iso timestamp: date +%Y%m%dT%H%M%S%z