Touch XML files to reflect date of original RRD file. Don't run dump in parallel...

[chef.git] / cookbooks / planet / templates / default / planetdump.erb

diff --git a/cookbooks/planet/templates/default/planetdump.erb b/cookbooks/planet/templates/default/planetdump.erb
index 8c9eb407b264f9ec4020cb286d288ccde1ea81c5..15a0051910152841cb635e898e8e7294b5e4d111 100644 (file)
--- a/cookbooks/planet/templates/default/planetdump.erb
+++ b/cookbooks/planet/templates/default/planetdump.erb
@@ -19,19 +19,37 @@ date="${year:2:2}${BASH_REMATCH[2]}${BASH_REMATCH[3]}"
 # Check the lock
 if [ -f /tmp/planetdump.lock ]; then
     if [ "$(ps -p `cat /tmp/planetdump.lock` | wc -l)" -gt 1 ]; then
-       echo "Error: Another planetdump is running"
-       exit 1
+        echo "Error: Another planetdump is running"
+        exit 1
     else
-       rm /tmp/planetdump.lock
+        rm /tmp/planetdump.lock
     fi
 fi
 
-# Create Lock
+# 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
 
 # Define cleanup function
 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
@@ -42,7 +60,7 @@ cd /store/planetdump
 
 # Cleanup
 rm -rf users
-rm -rf changesets changeset_tags
+rm -rf changesets changeset_tags changeset_comments
 rm -rf nodes node_tags
 rm -rf ways way_tags way_nodes
 rm -rf relations relation_tags relation_members
@@ -51,18 +69,34 @@ rm -rf relations relation_tags relation_members
 time nice -n 19 /opt/planet-dump-ng/planet-dump-ng \
      -c "pbzip2 -c" -f "/store/backup/${file}" --dense-nodes=1 \
      -C "changesets-${date}.osm.bz2" \
+     -D "discussions-${date}.osm.bz2" \
      -x "planet-${date}.osm.bz2" -X "history-${date}.osm.bz2" \
      -p "planet-${date}.osm.pbf" -P "history-${date}.osm.pbf"
 
-# Move XML dumps into place
-<% {
-    "changesets-${date}.osm.bz2" => "#{node[:planet][:dump][:xml_directory]}/${year}",
-    "planet-${date}.osm.bz2" => "#{node[:planet][:dump][:xml_directory]}/${year}",
-    "history-${date}.osm.bz2" => "#{node[:planet][:dump][:xml_history_directory]}/${year}",
-    "planet-${date}.osm.pbf" => node[:planet][:dump][:pbf_directory],
-    "history-${date}.osm.pbf" => node[:planet][:dump][:pbf_history_directory]
-}.each do |file,dir| -%>
-md5sum "<%= file %>" > "<%= file %>.md5"
-mkdir -p "<%= dir %>"
-mv "<%= file %>" "<%= file %>.md5" "<%= dir %>"
-<% end -%>
+# 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}"
+
+  md5sum "${name}" > "${name}.md5"
+  mkdir -p "${dir}/${year}"
+  mv "${name}" "${name}.md5" "${dir}/${year}"
+  ln -sf "${year:-.}/${name}" "${dir}/${latest}"
+  rm -f "${dir}/${latest}.md5"
+  sed -e "s/${name}/${latest}/" "${dir}/${year}/${name}.md5" > "${dir}/${latest}.md5"
+}
+
+# Move dumps into place
+install_dump "changesets" "bz2" "<%= node[:planet][:dump][:xml_directory] %>" "${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