cookbooks/planet/templates/default/planetdump.erb

   1 #!/bin/bash
   2
   3 # DO NOT EDIT - This file is being maintained by Chef
   4
   5 # Exit on error
   6 set -e
   7
   8 # Get the name of the file and the expected pattern
   9 file="$1"
  10 pattern="^osm-([0-9]{4})-([0-9]{2})-([0-9]{2})\.dmp$"
  11
  12 # Give up now if the file isn't a database dump
  13 [[ $file =~ $pattern ]] || exit 0
  14
  15 # Save the year and date from the file name
  16 year="${BASH_REMATCH[1]}"
  17 date="${year:2:2}${BASH_REMATCH[2]}${BASH_REMATCH[3]}"
  18
  19 # Check the lock
  20 if [ -f /tmp/planetdump.lock ]; then
  21     if [ "$(ps -p `cat /tmp/planetdump.lock` | wc -l)" -gt 1 ]; then
  22         echo "Error: Another planetdump is running"
  23         exit 1
  24     else
  25         rm /tmp/planetdump.lock
  26     fi
  27 fi
  28
  29 # Create lock file
  30 echo $$ > /tmp/planetdump.lock
  31
  32 # Define cleanup function
  33 function cleanup {
  34     # Remove the lock file
  35     rm /tmp/planetdump.lock
  36 }
  37
  38 # Remove lock on exit
  39 trap cleanup EXIT
  40
  41 # Change to working directory
  42 cd /store/planetdump
  43
  44 # Cleanup
  45 rm -rf users
  46 rm -rf changesets changeset_tags changeset_comments
  47 rm -rf nodes node_tags
  48 rm -rf ways way_tags way_nodes
  49 rm -rf relations relation_tags relation_members
  50
  51 # Run the dump
  52 time nice -n 19 /opt/planet-dump-ng/planet-dump-ng \
  53      --max-concurrency=4 \
  54      -c "pbzip2 -c" -f "/store/backup/${file}" --dense-nodes=1 \
  55      -C "changesets-${date}.osm.bz2" \
  56      -D "discussions-${date}.osm.bz2" \
  57      -x "planet-${date}.osm.bz2" -X "history-${date}.osm.bz2" \
  58      -p "planet-${date}.osm.pbf" -P "history-${date}.osm.pbf"
  59
  60 # Function to create bittorrent files
  61 function mk_torrent {
  62   local type="$1"
  63   local format="$2"
  64   local dir="$3"
  65   local s3path="$4"
  66   local s_year="$5"
  67   local web_dir="${dir}${s_year}"
  68   local name="${type}-${date}.osm.${format}"
  69   local web_path="${web_dir}/${name}"
  70   local s3_web_path="${s3path}/${name}"
  71   local rss_web_dir="https://planet.openstreetmap.org/${dir}"
  72   local rss_file="${type}-${format}-rss.xml"
  73   local torrent_file="${name}.torrent"
  74   local torrent_url="${rss_web_dir}${s_year}/${torrent_file}"
  75
  76   # create .torrent file
  77   mktorrent -l 22 "${name}" \
  78      -a udp://tracker.opentrackr.org:1337 \
  79      -a udp://tracker.datacenterlight.ch:6969/announce,http://tracker.datacenterlight.ch:6969/announce \
  80      -a udp://tracker.torrent.eu.org:451 \
  81      -a udp://tracker-udp.gbitt.info:80/announce,http://tracker.gbitt.info/announce,https://tracker.gbitt.info/announce \
  82      -a http://retracker.local/announce \
  83      -w "https://planet.openstreetmap.org/${web_path}" \
  84      -w "https://osm-planet-eu-central-1.s3.dualstack.eu-central-1.amazonaws.com/${s3_web_path}" \
  85      -w "https://osm-planet-us-west-2.s3.dualstack.us-west-2.amazonaws.com/${s3_web_path}" \
  86      -w "https://ftp5.gwdg.de/pub/misc/openstreetmap/planet.openstreetmap.org/${web_path}" \
  87      -w "https://ftpmirror.your.org/pub/openstreetmap/${web_path}" \
  88      -w "https://mirror.init7.net/openstreetmap/${web_path}" \
  89      -w "https://ftp.fau.de/osm-planet/${web_path}" \
  90      -w "https://ftp.spline.de/pub/openstreetmap/${web_path}" \
  91      -w "https://downloads.opencagedata.com/planet/${name}" \
  92      -w "https://planet.osm-hr.org/${web_path}" \
  93      -w "https://planet.maps.mail.ru/${web_path}" \
  94      -c "OpenStreetMap ${type} data export, licensed under https://opendatacommons.org/licenses/odbl/ by OpenStreetMap contributors" \
  95      -o "${torrent_file}" > /dev/null
  96
  97   # create .xml global RSS headers if missing
  98   torrent_time_rfc="$(date -R -r ${torrent_file})"
  99   test -f "${rss_file}" || echo "<x/>" | xmlstarlet select --xml-decl --indent \
 100         -N "atom=http://www.w3.org/2005/Atom" \
 101         -N "dcterms=http://purl.org/dc/terms/" \
 102         -N "content=http://purl.org/rss/1.0/modules/content/" \
 103         --encode "UTF-8" \
 104         --template \
 105         --match / \
 106         --elem "rss" \
 107                 --attr "version" --output "2.0" --break \
 108                 --attr "atom:DUMMY" --break \
 109         --elem "channel" \
 110         --elem "title" --output "OpenStreetMap ${type} ${format} torrent RSS" --break \
 111         --elem "link"  --output "${rss_web_dir}" --break \
 112         --elem "atom:link" \
 113                 --attr "href" --output "${rss_web_dir}/${rss_file}" --break \
 114                 --attr "rel" --output "self" --break \
 115                 --attr "type" --output "application/rss+xml" --break \
 116                 --break \
 117         --elem "description" --output "${type}.osm.${format}.torrent RSS feed" --break \
 118         --elem "copyright" --output "Source: OpenStreetMap contributors, under ODbL 1.0 licence" --break \
 119         --elem "generator" --output "OpenStreetMap xmlstarlet powered shell script v1.0" --break \
 120         --elem "language" --output "en" --break \
 121         --elem "lastBuildDate" --output "${torrent_time_rfc}" \
 122         > "${rss_file}"
 123
 124   # add newly created .torrent file as new entry to .xml RSS feed, removing excess entries
 125   torrent_size="$(stat --format="%s" ${torrent_file})"
 126   xmlstarlet edit --inplace \
 127         -a "//lastBuildDate" -t elem -n item -v ""  \
 128         -s "//item[1]" -t elem -n "title" -v "${torrent_file}" \
 129         -s "//item[1]" -t elem -n "guid" -v "${torrent_url}" \
 130         -s "//item[1]" -t elem -n "link" -v "${torrent_url}" \
 131         -s "//item[1]" -t elem -n "pubDate" -v "${torrent_time_rfc}" \
 132         -s "//item[1]" -t elem -n "category" -v "OpenStreetMap data" \
 133         -s "//item[1]" -t elem -n "enclosure" \
 134                 -s "//item[1]"/enclosure -t attr -n "type" -v "application/x-bittorrent" \
 135                 -s "//item[1]"/enclosure -t attr -n "length" -v "${torrent_size}" \
 136                 -s "//item[1]"/enclosure -t attr -n "url" -v "${torrent_url}" \
 137         -s "//item[1]" -t elem -n "description" -v "OpenStreetMap torrent ${torrent_file}" \
 138         -u /rss/channel/lastBuildDate -v "${torrent_time_rfc}" \
 139         -d /rss/@atom:DUMMY \
 140         -d "//item[position()>5]" \
 141         "${rss_file}"
 142 }
 143
 144 function replication_status_wait {
 145   local s3_url="$1"
 146   for i in {1..3600}; do
 147     local replication_status=$(curl -sI --location "${s3_url}" | grep -F 'x-amz-replication-status' | awk '{print $2}' |  tr -d '\r' )
 148
 149     if [[ "${replication_status}" == "COMPLETED" ]]; then
 150       return 0 # success
 151     fi
 152
 153     sleep 1
 154   done
 155   echo "Timeout waiting for ${s3_url} to complete replication status: ${replication_status}"
 156 }
 157
 158 # Function to install a dump in place
 159 function install_dump {
 160   local type="$1"
 161   local format="$2"
 162   local dir="$3"
 163   local s3dir="$4"
 164   local year="$5"
 165   local name="${type}-${date}.osm.${format}"
 166   local latest="${type}-latest.osm.${format}"
 167   local rss_file="${type}-${format}-rss.xml"
 168
 169   md5sum "${name}" > "${name}.md5"
 170
 171   # Upload all files to S3
 172   /opt/awscli/v2/current/bin/aws --profile=osm-pds-upload s3 cp --storage-class=INTELLIGENT_TIERING --no-progress "${name}.md5" "s3://osm-planet-eu-central-1/${s3dir}/${name}.md5"
 173   /opt/awscli/v2/current/bin/aws --profile=osm-pds-upload s3 cp --storage-class=INTELLIGENT_TIERING --no-progress "${name}.torrent" "s3://osm-planet-eu-central-1/${s3dir}/${name}.torrent"
 174   /opt/awscli/v2/current/bin/aws --profile=osm-pds-upload s3 cp --storage-class=INTELLIGENT_TIERING --no-progress "${name}" "s3://osm-planet-eu-central-1/${s3dir}/${name}"
 175
 176   # Waiting for S3 replication to complete
 177   replication_status_wait "https://osm-planet-eu-central-1.s3.dualstack.eu-central-1.amazonaws.com/${s3dir}/${name}.md5"
 178   replication_status_wait "https://osm-planet-eu-central-1.s3.dualstack.eu-central-1.amazonaws.com/${s3dir}/${name}.torrent"
 179   replication_status_wait "https://osm-planet-eu-central-1.s3.dualstack.eu-central-1.amazonaws.com/${s3dir}/${name}"
 180
 181   mkdir -p "${dir}/${year}"
 182   mv "${name}" "${name}.md5" "${dir}/${year}"
 183   ln -sf "${year:-.}/${name}" "${dir}/${latest}"
 184   test -f "${name}.torrent" && mv "${name}.torrent" "${dir}/${year}" && ln -sf "${year:-.}/${name}.torrent" "${dir}/${latest}.torrent"
 185   test -f "${rss_file}" && xmllint --noout "${rss_file}" && cp -f "${rss_file}" "${dir}"
 186   rm -f "${dir}/${latest}.md5"
 187   sed -e "s/${name}/${latest}/" "${dir}/${year}/${name}.md5" > "${dir}/${latest}.md5"
 188 }
 189
 190 # Create *.torrent files
 191 mk_torrent "changesets" "bz2" "planet" "changesets/osm/${year}" "/${year}"
 192 mk_torrent "discussions" "bz2" "planet" "discussions/osm/${year}" "/${year}"
 193 mk_torrent "planet" "bz2" "planet" "planet/osm/${year}" "/${year}"
 194 mk_torrent "history" "bz2" "planet/full-history" "planet-full-history/osm/${year}" "/${year}"
 195 mk_torrent "planet" "pbf" "pbf" "planet/pbf/${year}"
 196 mk_torrent "history" "pbf" "pbf/full-history" "planet-full-history/pbf/${year}"
 197
 198 # Move dumps into place
 199 install_dump "changesets" "bz2" "<%= node[:planet][:dump][:xml_directory] %>" "changesets/osm/${year}" "${year}"
 200 install_dump "discussions" "bz2" "<%= node[:planet][:dump][:xml_directory] %>" "discussions/osm/${year}" "${year}"
 201 install_dump "planet" "bz2" "<%= node[:planet][:dump][:xml_directory] %>" "planet/osm/${year}" "${year}"
 202 install_dump "history" "bz2" "<%= node[:planet][:dump][:xml_history_directory] %>" "planet-full-history/osm/${year}" "${year}"
 203 install_dump "planet" "pbf" "<%= node[:planet][:dump][:pbf_directory] %>" "planet/pbf/${year}"
 204 install_dump "history" "pbf" "<%= node[:planet][:dump][:pbf_history_directory] %>" "planet-full-history/pbf/${year}"