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
  25 usage() {
  26   cat <<'EOF'
  27 btrbk-run [OPTIONS]
  28 usually -t TARGET_HOST or -s SOURCE_HOST
  29
  30 Note, at source location, intentionally not executable, run and read
  31 install-my-scripts.
  32
  33 EOF
  34   echo "top of script file:"
  35   sed -n '1,/^[# ]*end command line/{p;b};q' "$0"
  36   exit $1
  37 }
  38
  39 rsync-dirs() {
  40   local host=$1
  41   local path=$2
  42   m rsync $dry_run_arg -ahi --relative --delete "$path" "root@$host:/"
  43 }
  44
  45
  46 m() { printf "%s: %s\n" "${0##*/}" "$*";  "$@"; }
  47
  48 # latest $MAIL_HOST
  49 if [[ -e /b/bash_unpublished/source-semi-priv ]]; then
  50   source /b/bash_unpublished/source-semi-priv
  51 fi
  52
  53 # note q is owned by root:1000
  54
  55 mountpoints=()
  56
  57 rsync_mountpoint=/q
  58
  59 # default options
  60 conf_only=false
  61 dry_run=false # mostly for testing
  62 rate_limit=no
  63 verbose=true; verbose_arg=-v
  64 progress_arg="--progress"
  65 pull_reexec=false
  66
  67 default_args_file=/etc/btrbk-run.conf
  68 if [[ -s $default_args_file ]]; then
  69   set -- $(< $default_args_file) "$@"
  70   echo "$0: warning: default btrbk-run options set in $default_args_file (sleeping 5 seconds):"
  71   cat $default_args_file
  72   sleep 5
  73 fi
  74
  75 orig_args=("$@")
  76 temp=$(getopt -l pull-reexec,help cl:m:nps:t:vh "$@") || usage 1
  77 eval set -- "$temp"
  78 while true; do
  79   case $1 in
  80     # only creates the config file, does not run btrbk
  81     -c) conf_only=true; shift ;;
  82     # bytes per second, suffix k m g
  83     -l) rate_limit=$2; shift 2 ;;
  84     # Comma separated mountpoints to backup. This has defaults set below.
  85     -m) IFS=, mountpoints=($2); unset IFS; shift 2 ;;
  86     -n) dry_run=true; dry_run_arg=-n; shift ;;
  87     -p) progress_arg="--progress"; shift ;;
  88     --pull-reexec) pull_reexec=true; shift ;;
  89     -q) verbose=false; verbose_arg=; progress_arg=; shift ;;
  90     # source host to receive a backup from
  91     -s) source=$2; shift 2 ;;
  92     # target hosts to send to. empty is valid for just doing local
  93     # snapshot. we have default hosts we will populate.
  94     -t) IFS=, targets=($2); unset IFS; shift 2 ;;
  95     -v) verbose=true; verbose_arg=-v; shift ;;
  96     -h|--help) usage ;;
  97     --) shift; break ;;
  98     *) echo "$0: Internal error!" ; exit 1 ;;
  99   esac
 100 done
 101
 102 # usefull commands are resume and archive
 103 cmd_arg=${1:-run}
 104
 105 if [[ -v targets && $source ]]; then
 106   echo "$0: error: -t and -s are mutually exclusive" >&2
 107   exit 1
 108 fi
 109
 110
 111 echo -e "$0: options: conf_only=$conf_only\ndry_run=$dry_run\nrate_limit=$rate_limit\nverbose=$verbose\ncmd_arg=$cmd_arg"
 112 ### end options parsing
 113
 114
 115 # set default targets
 116 if [[ ! -v targets && ! $source ]]; then
 117   case $HOSTNAME in
 118     x2|kw)
 119       if [[ $HOSTNAME == "$MAIL_HOST" ]]; then
 120         targets=($HOME_DOMAIN)
 121       fi
 122       ;;
 123     tp)
 124       targets=(frodo)
 125       if [[ $HOSTNAME == "$MAIL_HOST" ]]; then
 126         if timeout -s 9 10 ssh x2 :; then
 127           targets+=(x2)
 128         fi
 129       fi
 130       ;;
 131     frodo)
 132       targets=()
 133       ;;
 134     *)
 135       echo "$0: error: no default targets for this host, use -t"
 136       exit 1
 137       ;;
 138   esac
 139 fi
 140
 141 if [[ -v targets ]]; then
 142   echo "targets: ${targets[*]}"
 143 fi
 144
 145 if [[ $source ]]; then
 146   echo "source: $source"
 147 fi
 148
 149
 150 if [[ $mountpoints ]]; then
 151   for mp in ${mountpoints[@]}; do # default mountpoints to sync
 152     if [[ -e /nocow/btrfs-stale/$mp ]]; then
 153       echo "$0: error: $mp is stale, mount-latest-subvol first"
 154       exit 1
 155     fi
 156   done
 157 else
 158   # set default mountpoints
 159   case $HOSTNAME in
 160     # no remote backups atm. note, if we do enable this, configuration below will need some changes.
 161     #  frodo)
 162     #    prospective_mps=(/i)
 163     #    ;;
 164     *)
 165       prospective_mps=(/a /q)
 166       if [[ $HOSTNAME == "$MAIL_HOST" ]]; then
 167         prospective_mps+=(/o)
 168       fi
 169       ;;
 170   esac
 171   for mp in ${prospective_mps[@]}; do # default mountpoints to sync
 172     if [[ -e /nocow/btrfs-stale/$mp ]]; then
 173       echo "$0: warning: $mp stale, not adding to default mountpoints"
 174       continue
 175     fi
 176     if awk '{print $2}' /etc/fstab | grep -xF $mp &>/dev/null; then
 177       mountpoints+=($mp)
 178     fi
 179   done
 180 fi
 181
 182 echo "mountpoints: ${mountpoints[*]}"
 183
 184
 185
 186 # pull_reexec stops us from getting into an infinite loop if there is some
 187 # kind of weird problem
 188 pulla=false
 189 for m in "${mountpoints[@]}"; do
 190   if [[ $m == /a ]]; then
 191     pulla=true
 192     break
 193   fi
 194 done
 195 if ! $pull_reexec && [[ $source ]] && ! $pulla ; then
 196   tmpf=$(mktemp)
 197   scp $source:/a/bin/distro-setup/btrbk-run $tmpf
 198   if ! diff -q $tmpf $BASH_SOURCE; then
 199     echo "$0: found newer version on host $source. reexecing"
 200     install -T $tmpf /usr/local/bin/btrbk-run
 201     m /usr/local/bin/btrbk-run --pull-reexec "${orig_args[@]}"
 202     exit
 203   fi
 204 fi
 205
 206
 207 ##### end command line parsing ########
 208
 209
 210
 211 if ! which btrbk &>/dev/null; then
 212   echo "$0: error: no btrbk binary found"
 213   exit 1
 214 fi
 215 # if our mountpoints are from stale snapshots,
 216 # it doesn't make sense to do a backup.
 217 check-subvol-stale ${mountpoints[@]} || exit 1
 218
 219
 220 cat >/etc/btrbk.conf <<EOF
 221 ssh_identity /root/.ssh/home
 222 # Just a guess that local7 is a good facility to pick.
 223 # It's a bit odd that the transaction log has to be logged to
 224 # a file or syslog, while other output is sent to std out.
 225 # The man does not mention a way for them to be together, but
 226 # I dunno if setting a log level like warn might also output
 227 # transaction info.
 228 transaction_syslog local7
 229
 230 # note, i had this because man said 20% speedup, but ran into
 231 # this issue, https://github.com/digint/btrbk/issues/275
 232 #stream_buffer 512m
 233
 234 # so we only run one at a time
 235 lockfile                   /var/lock/btrbk.lock
 236
 237 # default format of short does not accomidate hourly preservation setting
 238 timestamp_format long-iso
 239
 240 # only make a snapshot if things have changed
 241 snapshot_create onchange
 242 # I could make this different from target_preserve,
 243 # if one disk had less space.
 244 # for now, keeping them equal.
 245 snapshot_preserve 36h 14d 8w 24m
 246 snapshot_preserve_min 4h
 247 snapshot_dir btrbk
 248
 249 # so, total backups = ~89
 250 target_preserve 36h 14d 8w 24m
 251 target_preserve_min 4h
 252
 253 # if something fails and it's not obvious, try doing
 254 # btrbk -l debug -v dryrun
 255
 256 rate_limit $rate_limit
 257 EOF
 258
 259
 260
 261 # for an initial run, btrbk requires the dir to exist.
 262 mkdir -p /mnt/root/btrbk
 263 local_zone=$(date +%z)
 264 for h in ${targets[@]} $source; do
 265   zone=$(ssh root@$h "mkdir -p /mnt/root/btrbk; date +%z")
 266   if [[ $zone != $local_zone ]]; then
 267     echo "error: error. dont confuse yourself with multiple time zones. $h has different timezone than localhost" >&2
 268     exit 1
 269   fi
 270 done
 271
 272
 273
 274 vol=/mnt/root
 275 for m in ${mountpoints[@]}; do
 276   sub=${m##*/}
 277   if [[ $source ]]; then
 278     cat >>/etc/btrbk.conf <<EOF
 279 volume ssh://$source$vol
 280 subvolume $sub
 281 target send-receive $vol/btrbk
 282 EOF
 283   else
 284     cat >>/etc/btrbk.conf <<EOF
 285 volume $vol
 286 subvolume $sub
 287 EOF
 288     for tg in ${targets[@]}; do
 289       cat >>/etc/btrbk.conf <<EOF
 290 target send-receive ssh://$tg$vol/btrbk
 291 EOF
 292     done
 293   fi
 294 done
 295
 296
 297 # todo: umount first to ensure we don't have any errors
 298 # todo: do some kill fuser stuff to make umount more reliable
 299
 300
 301 if $conf_only; then
 302   exit
 303 fi
 304
 305
 306
 307 if $dry_run; then
 308   m btrbk -v -n $cmd_arg
 309   exit 0
 310 elif [[ $cmd_arg == archive ]]; then
 311   if [[ $source ]]; then
 312     m btrbk $verbose_arg $progress_arg $cmd_arg ssh://$source$vol $vol
 313   else
 314     for tg in ${targets[@]}; do
 315       m btrbk $verbose_arg $progress_arg $cmd_arg $vol ssh://$tg$vol
 316     done
 317   fi
 318   exit 0
 319 fi
 320 # -q and just using the syslog option seemed nice,
 321 # but it doesn't show when a send has a parent and when it doesn't.
 322 m btrbk $verbose_arg $progress_arg $cmd_arg
 323
 324 # if we have it, sync to systems which don't
 325 if mountpoint $rsync_mountpoint >/dev/null; then
 326   for tg in ${targets[@]}; do
 327     case $tg in
 328       li|lk)
 329         for x in /p/c/machine_specific/*.hosts; do
 330           if grep -qxF $tg $x; then
 331             dir=${x%.hosts}
 332             rsync-dirs $tg $dir
 333           fi
 334         done
 335         ;;
 336     esac
 337   done
 338 fi
 339
 340 /a/bin/distro-setup/install-my-scripts
 341 if [[ $source ]]; then
 342   m mount-latest-subvol
 343 else
 344   m /a/exe/mount-latest-remote ${targets[@]}
 345 fi
 346
 347 # todo: move variable data we don't care about backing up
 348 # to /nocow and symlink it.
 349
 350
 351 # background on btrbk timezones. with short/long, timestamps use local time.
 352 # for long, if your local time moves backwards, by moving timezones or
 353 # for an hour when daylight savings changes it, you will temporarily get
 354 # a more aggressive retention policy for the overlapping period, and
 355 # vice versa for the opposite timezone move. The alternative is using
 356 # long-iso, which puts timezone info into the timestamp, which means
 357 # that instead of shifting time, you shift the start of day/week/month
 358 # which is used for retention to your new local time, which means for
 359 # example, if you moved forward by 8 hours, the daily/weekly/monthly
 360 # retention will be 8 hours more aggressive since midnight is at a new
 361 # time, unless you fake the timzeone using the TZ env variable.
 362 # However, in the short term, there will be no inconsistencies.
 363 # I don't see any problem with shifting when the day starts for
 364 # retention, so I'm using long-iso.
 365
 366 # note to create a long-iso timestamp: date +%Y%m%dT%H%M%S%z