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