]> git.openstreetmap.org Git - chef.git/blobdiff - cookbooks/planet/templates/default/planetdump.erb
planet: add S3 buckets to torrents
[chef.git] / cookbooks / planet / templates / default / planetdump.erb
index ae6302382040f782704ba364f99dca154d1163cf..8ebb1239f4775f25dd838ea8556de1898b831e6c 100644 (file)
@@ -26,13 +26,6 @@ if [ -f /tmp/planetdump.lock ]; then
     fi
 fi
 
-# Redirect this shell's output to a file. This is so that it
-# can be emailed later, since this script is run from incron
-# and incron doesn't yet support MAILTO like cron does. The
-# command below appears to work in bash as well as dash.
-logfile="/tmp/planetdump.log.$$"
-exec > "${logfile}" 2>&1
-
 # Create lock file
 echo $$ > /tmp/planetdump.lock
 
@@ -40,16 +33,6 @@ echo $$ > /tmp/planetdump.lock
 function cleanup {
     # Remove the lock file
     rm /tmp/planetdump.lock
-
-    # Send an email with the output, since incron doesn't yet
-    # support doing this in the incrontab
-    if [[ -s "$logfile" ]]
-    then
-        mailx -s "Planet dump output: ${file}" zerebubuth@gmail.com < "${logfile}"
-    fi
-
-    # Remove the log file
-    rm -f "${logfile}"
 }
 
 # Remove lock on exit
@@ -76,72 +59,127 @@ time nice -n 19 /opt/planet-dump-ng/planet-dump-ng \
 
 # Function to create bittorrent files
 function mk_torrent {
-  type="$1"
-  format="$2"
-  web_dir="$3"
-  name="${type}-${date}.osm.${format}"
-  web_path="${web_dir}/${name}"
-
-  mktorrent -l 22 ${name} \
+  local type="$1"
+  local format="$2"
+  local dir="$3"
+  local s3path="$4"
+  local s_year="$5"
+  local web_dir="${dir}${s_year}"
+  local name="${type}-${date}.osm.${format}"
+  local web_path="${web_dir}/${name}"
+  local s3_web_path="${s3path}/${name}"
+  local rss_web_dir="https://planet.openstreetmap.org/${dir}"
+  local rss_file="${type}-${format}-rss.xml"
+  local torrent_file="${name}.torrent"
+  local torrent_url="${rss_web_dir}${s_year}/${torrent_file}"
+
+  # create .torrent file
+  mktorrent -l 22 "${name}" \
      -a udp://tracker.opentrackr.org:1337 \
      -a udp://tracker.datacenterlight.ch:6969/announce,http://tracker.datacenterlight.ch:6969/announce \
      -a udp://tracker.torrent.eu.org:451 \
      -a udp://tracker-udp.gbitt.info:80/announce,http://tracker.gbitt.info/announce,https://tracker.gbitt.info/announce \
      -a http://retracker.local/announce \
-     -w https://planet.openstreetmap.org/${web_path} \
-     -w https://ftp5.gwdg.de/pub/misc/openstreetmap/planet.openstreetmap.org/${web_path} \
-     -w https://ftpmirror.your.org/pub/openstreetmap/${web_path} \
-     -w https://mirror.init7.net/openstreetmap/${web_path} \
-     -w https://free.nchc.org.tw/osm.planet/${web_path} \
-     -w https://ftp.fau.de/osm-planet/${web_path} \
-     -w https://ftp.spline.de/pub/openstreetmap/${web_path} \
-     -w https://osm.openarchive.site/${name} \
-     -w https://downloads.opencagedata.com/planet/${name} \
+     -w "https://planet.openstreetmap.org/${web_path}" \
+     -w "https://osm-planet-eu-central-1.s3.dualstack.eu-central-1.amazonaws.com/${s3_web_path}" \
+     -w "https://osm-planet-us-west-2.s3.dualstack.us-west-2.amazonaws.com/${s3_web_path}" \
+     -w "https://ftp5.gwdg.de/pub/misc/openstreetmap/planet.openstreetmap.org/${web_path}" \
+     -w "https://ftpmirror.your.org/pub/openstreetmap/${web_path}" \
+     -w "https://mirror.init7.net/openstreetmap/${web_path}" \
+     -w "https://ftp.fau.de/osm-planet/${web_path}" \
+     -w "https://ftp.spline.de/pub/openstreetmap/${web_path}" \
+     -w "https://downloads.opencagedata.com/planet/${name}" \
+     -w "https://planet.osm-hr.org/${web_path}" \
+     -w "https://planet.maps.mail.ru/${web_path}" \
      -c "OpenStreetMap ${type} data export, licensed under https://opendatacommons.org/licenses/odbl/ by OpenStreetMap contributors" \
-     -o ${name}.torrent
+     -o "${torrent_file}" > /dev/null
+
+  # create .xml global RSS headers if missing
+  torrent_time_rfc="$(date -R -r ${torrent_file})"
+  test -f "${rss_file}" || echo "<x/>" | xmlstarlet select --xml-decl --indent \
+       -N "atom=http://www.w3.org/2005/Atom" \
+       -N "dcterms=http://purl.org/dc/terms/" \
+       -N "content=http://purl.org/rss/1.0/modules/content/" \
+       --encode "UTF-8" \
+       --template \
+       --match / \
+       --elem "rss" \
+               --attr "version" --output "2.0" --break \
+               --attr "atom:DUMMY" --break \
+       --elem "channel" \
+       --elem "title" --output "OpenStreetMap ${type} ${format} torrent RSS" --break \
+       --elem "link"  --output "${rss_web_dir}" --break \
+       --elem "atom:link" \
+               --attr "href" --output "${rss_web_dir}/${rss_file}" --break \
+               --attr "rel" --output "self" --break \
+               --attr "type" --output "application/rss+xml" --break \
+               --break \
+       --elem "description" --output "${type}.osm.${format}.torrent RSS feed" --break \
+       --elem "copyright" --output "Source: OpenStreetMap contributors, under ODbL 1.0 licence" --break \
+       --elem "generator" --output "OpenStreetMap xmlstarlet powered shell script v1.0" --break \
+       --elem "language" --output "en" --break \
+       --elem "lastBuildDate" --output "${torrent_time_rfc}" \
+       > "${rss_file}"
+
+  # add newly created .torrent file as new entry to .xml RSS feed, removing excess entries
+  torrent_size="$(stat --format="%s" ${torrent_file})"
+  xmlstarlet edit --inplace \
+       -a "//lastBuildDate" -t elem -n item -v ""  \
+       -s "//item[1]" -t elem -n "title" -v "${torrent_file}" \
+       -s "//item[1]" -t elem -n "guid" -v "${torrent_url}" \
+       -s "//item[1]" -t elem -n "link" -v "${torrent_url}" \
+       -s "//item[1]" -t elem -n "pubDate" -v "${torrent_time_rfc}" \
+       -s "//item[1]" -t elem -n "category" -v "OpenStreetMap data" \
+       -s "//item[1]" -t elem -n "enclosure" \
+               -s "//item[1]"/enclosure -t attr -n "type" -v "application/x-bittorrent" \
+               -s "//item[1]"/enclosure -t attr -n "length" -v "${torrent_size}" \
+               -s "//item[1]"/enclosure -t attr -n "url" -v "${torrent_url}" \
+       -s "//item[1]" -t elem -n "description" -v "OpenStreetMap torrent ${torrent_file}" \
+       -u /rss/channel/lastBuildDate -v "${torrent_time_rfc}" \
+       -d /rss/@atom:DUMMY \
+       -d "//item[position()>5]" \
+       "${rss_file}"
 }
 
 # Function to install a dump in place
 function install_dump {
-  type="$1"
-  format="$2"
-  dir="$3"
-  year="$4"
-  name="${type}-${date}.osm.${format}"
-  latest="${type}-latest.osm.${format}"
+  local type="$1"
+  local format="$2"
+  local dir="$3"
+  local s3dir="$4"
+  local year="$5"
+  local name="${type}-${date}.osm.${format}"
+  local latest="${type}-latest.osm.${format}"
+  local rss_file="${type}-${format}-rss.xml"
 
   md5sum "${name}" > "${name}.md5"
+
+  # Upload all files to S3
+  /opt/awscli/v2/current/bin/aws --profile=osm-pds-upload s3 cp --storage-class=INTELLIGENT_TIERING --no-progress "${name}.md5" "${s3dir}/${name}.md5"
+  /opt/awscli/v2/current/bin/aws --profile=osm-pds-upload s3 cp --storage-class=INTELLIGENT_TIERING --no-progress "${name}.torrent" "${s3dir}/${name}.torrent"
+  /opt/awscli/v2/current/bin/aws --profile=osm-pds-upload s3 cp --storage-class=INTELLIGENT_TIERING --no-progress "${name}" "${s3dir}/${name}"
+
   mkdir -p "${dir}/${year}"
-  test -f "${name}.torrent" && mv "${name}.torrent" "${dir}/${year}"
   mv "${name}" "${name}.md5" "${dir}/${year}"
   ln -sf "${year:-.}/${name}" "${dir}/${latest}"
+  test -f "${name}.torrent" && mv "${name}.torrent" "${dir}/${year}" && ln -sf "${year:-.}/${name}.torrent" "${dir}/${latest}.torrent"
+  test -f "${rss_file}" && xmllint --noout "${rss_file}" && cp -f "${rss_file}" "${dir}"
   rm -f "${dir}/${latest}.md5"
   sed -e "s/${name}/${latest}/" "${dir}/${year}/${name}.md5" > "${dir}/${latest}.md5"
 }
 
 # Create *.torrent files
-mk_torrent "changesets" "bz2" "planet/${year}"
-mk_torrent "discussions" "bz2" "planet/${year}"
-mk_torrent "planet" "bz2" "planet/${year}"
-mk_torrent "history" "bz2" "planet/full-history/${year}"
-mk_torrent "planet" "pbf" "pbf"
-mk_torrent "history" "pbf" "pbf/full-history"
+mk_torrent "changesets" "bz2" "planet" "changesets/osm/${year}" "/${year}"
+mk_torrent "discussions" "bz2" "planet" "discussions/osm/${year}" "/${year}"
+mk_torrent "planet" "bz2" "planet" "planet/osm/${year}" "/${year}"
+mk_torrent "history" "bz2" "planet/full-history" "planet-full-history/osm/${year}" "/${year}"
+mk_torrent "planet" "pbf" "pbf" "planet/pbf/${year}"
+mk_torrent "history" "pbf" "pbf/full-history" "planet-full-history/pbf/${year}"
 
 # Move dumps into place
-install_dump "changesets" "bz2" "<%= node[:planet][:dump][:xml_directory] %>" "${year}"
-install_dump "discussions" "bz2" "<%= node[:planet][:dump][:xml_directory] %>" "${year}"
-install_dump "planet" "bz2" "<%= node[:planet][:dump][:xml_directory] %>" "${year}"
-install_dump "history" "bz2" "<%= node[:planet][:dump][:xml_history_directory] %>" "${year}"
-install_dump "planet" "pbf" "<%= node[:planet][:dump][:pbf_directory] %>"
-install_dump "history" "pbf" "<%= node[:planet][:dump][:pbf_history_directory] %>"
-
-# Remove pbf dumps older than 90 days
-find "<%= node[:planet][:dump][:pbf_directory] %>" "<%= node[:planet][:dump][:pbf_history_directory] %>" \
-     -maxdepth 1 -mindepth 1 -type f -mtime +90 \
-     \( \
-     -iname 'planet-*.pbf' \
-     -o -iname 'history-*.pbf' \
-     -o -iname 'planet-*.pbf.md5' \
-     -o -iname 'history-*.pbf.md5' \
-     \) \
-     -delete
+install_dump "changesets" "bz2" "<%= node[:planet][:dump][:xml_directory] %>" "s3://osm-planet-eu-central-1/changesets/osm/${year}" "${year}"
+install_dump "discussions" "bz2" "<%= node[:planet][:dump][:xml_directory] %>" "s3://osm-planet-eu-central-1/discussions/osm/${year}" "${year}"
+install_dump "planet" "bz2" "<%= node[:planet][:dump][:xml_directory] %>" "s3://osm-planet-eu-central-1/planet/osm/${year}" "${year}"
+install_dump "history" "bz2" "<%= node[:planet][:dump][:xml_history_directory] %>" "s3://osm-planet-eu-central-1/planet-full-history/osm/${year}" "${year}"
+install_dump "planet" "pbf" "<%= node[:planet][:dump][:pbf_directory] %>" "s3://osm-planet-eu-central-1/planet/pbf/${year}"
+install_dump "history" "pbf" "<%= node[:planet][:dump][:pbf_history_directory] %>" "s3://osm-planet-eu-central-1/planet-full-history/pbf/${year}"