Merge pull request #362 from hbogner/master
[chef.git] / 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 # Redirect this shell's output to a file. This is so that it
30 # can be emailed later, since this script is run from incron
31 # and incron doesn't yet support MAILTO like cron does. The
32 # command below appears to work in bash as well as dash.
33 logfile="/tmp/planetdump.log.$$"
34 exec > "${logfile}" 2>&1
35
36 # Create lock file
37 echo $$ > /tmp/planetdump.lock
38
39 # Define cleanup function
40 function cleanup {
41     # Remove the lock file
42     rm /tmp/planetdump.lock
43
44     # Send an email with the output, since incron doesn't yet
45     # support doing this in the incrontab
46     if [[ -s "$logfile" ]]
47     then
48         mailx -s "Planet dump output: ${file}" zerebubuth@gmail.com < "${logfile}"
49     fi
50
51     # Remove the log file
52     rm -f "${logfile}"
53 }
54
55 # Remove lock on exit
56 trap cleanup EXIT
57
58 # Change to working directory
59 cd /store/planetdump
60
61 # Cleanup
62 rm -rf users
63 rm -rf changesets changeset_tags changeset_comments
64 rm -rf nodes node_tags
65 rm -rf ways way_tags way_nodes
66 rm -rf relations relation_tags relation_members
67
68 # Run the dump
69 time nice -n 19 /opt/planet-dump-ng/planet-dump-ng \
70      --max-concurrency=4 \
71      -c "pbzip2 -c" -f "/store/backup/${file}" --dense-nodes=1 \
72      -C "changesets-${date}.osm.bz2" \
73      -D "discussions-${date}.osm.bz2" \
74      -x "planet-${date}.osm.bz2" -X "history-${date}.osm.bz2" \
75      -p "planet-${date}.osm.pbf" -P "history-${date}.osm.pbf"
76
77 # Function to create bittorrent files
78 function mk_torrent {
79   type="$1"
80   format="$2"
81   web_dir="$3"
82   name="${type}-${date}.osm.${format}"
83   web_path="${web_dir}/${name}"
84
85   mktorrent -l 22 ${name} \
86      -a udp://tracker.opentrackr.org:1337 \
87      -a udp://tracker.datacenterlight.ch:6969/announce,http://tracker.datacenterlight.ch:6969/announce \
88      -a udp://tracker.torrent.eu.org:451 \
89      -a udp://tracker-udp.gbitt.info:80/announce,http://tracker.gbitt.info/announce,https://tracker.gbitt.info/announce \
90      -a http://retracker.local/announce \
91      -w https://planet.openstreetmap.org/${web_path} \
92      -w https://ftp5.gwdg.de/pub/misc/openstreetmap/planet.openstreetmap.org/${web_path} \
93      -w https://ftpmirror.your.org/pub/openstreetmap/${web_path} \
94      -w https://mirror.init7.net/openstreetmap/${web_path} \
95      -w https://free.nchc.org.tw/osm.planet/${web_path} \
96      -w https://ftp.fau.de/osm-planet/${web_path} \
97      -w https://ftp.spline.de/pub/openstreetmap/${web_path} \
98      -w https://osm.openarchive.site/${name} \
99      -w https://downloads.opencagedata.com/planet/${name} \
100      -w https://planet.osm-hr.org/${web_path} \
101      -c "OpenStreetMap ${type} data export, licensed under https://opendatacommons.org/licenses/odbl/ by OpenStreetMap contributors" \
102      -o ${name}.torrent
103 }
104
105 # Function to install a dump in place
106 function install_dump {
107   type="$1"
108   format="$2"
109   dir="$3"
110   year="$4"
111   name="${type}-${date}.osm.${format}"
112   latest="${type}-latest.osm.${format}"
113
114   md5sum "${name}" > "${name}.md5"
115   mkdir -p "${dir}/${year}"
116   test -f "${name}.torrent" && mv "${name}.torrent" "${dir}/${year}"
117   mv "${name}" "${name}.md5" "${dir}/${year}"
118   ln -sf "${year:-.}/${name}" "${dir}/${latest}"
119   rm -f "${dir}/${latest}.md5"
120   sed -e "s/${name}/${latest}/" "${dir}/${year}/${name}.md5" > "${dir}/${latest}.md5"
121 }
122
123 # Create *.torrent files
124 mk_torrent "changesets" "bz2" "planet/${year}"
125 mk_torrent "discussions" "bz2" "planet/${year}"
126 mk_torrent "planet" "bz2" "planet/${year}"
127 mk_torrent "history" "bz2" "planet/full-history/${year}"
128 mk_torrent "planet" "pbf" "pbf"
129 mk_torrent "history" "pbf" "pbf/full-history"
130
131 # Move dumps into place
132 install_dump "changesets" "bz2" "<%= node[:planet][:dump][:xml_directory] %>" "${year}"
133 install_dump "discussions" "bz2" "<%= node[:planet][:dump][:xml_directory] %>" "${year}"
134 install_dump "planet" "bz2" "<%= node[:planet][:dump][:xml_directory] %>" "${year}"
135 install_dump "history" "bz2" "<%= node[:planet][:dump][:xml_history_directory] %>" "${year}"
136 install_dump "planet" "pbf" "<%= node[:planet][:dump][:pbf_directory] %>"
137 install_dump "history" "pbf" "<%= node[:planet][:dump][:pbf_history_directory] %>"
138
139 # Remove pbf dumps older than 90 days
140 find "<%= node[:planet][:dump][:pbf_directory] %>" "<%= node[:planet][:dump][:pbf_history_directory] %>" \
141      -maxdepth 1 -mindepth 1 -type f -mtime +90 \
142      \( \
143      -iname 'planet-*.pbf' \
144      -o -iname 'history-*.pbf' \
145      -o -iname 'planet-*.pbf.md5' \
146      -o -iname 'history-*.pbf.md5' \
147      \) \
148      -delete