Copy changeset replication diffs to S3

[chef.git] / cookbooks / planet / files / default / replication-bin / replicate-changesets

diff --git a/cookbooks/planet/files/default/replication-bin/replicate-changesets b/cookbooks/planet/files/default/replication-bin/replicate-changesets
index 3cf54291ff26eb4305eca4ee3d40b5d70f6fda6f..87334f21a06b51f840b01e708a6a03c4fce48a71 100755 (executable)
--- a/cookbooks/planet/files/default/replication-bin/replicate-changesets
+++ b/cookbooks/planet/files/default/replication-bin/replicate-changesets
@@ -143,7 +143,7 @@ class Replicator
   def initialize(config)
     @config = YAML.safe_load(File.read(config))
     @state = YAML.safe_load(File.read(@config["state_file"]), [Time])
-    @conn = PGconn.connect(@config["db"])
+    @conn = PG::Connection.connect(@config["db"])
     # get current time from the database rather than the current system
     @now = @conn.exec("select now() as now").map { |row| Time.parse(row["now"]) }[0]
   end
@@ -187,8 +187,8 @@ class Replicator
     { "version" => "0.6",
       "generator" => "replicate_changesets.rb",
       "copyright" => "OpenStreetMap and contributors",
-      "attribution" => "http://www.openstreetmap.org/copyright",
-      "license" => "http://opendatacommons.org/licenses/odbl/1-0/" }
+      "attribution" => "https://www.openstreetmap.org/copyright",
+      "license" => "https://opendatacommons.org/licenses/odbl/1-0/" }
       .each { |k, v| doc.root[k] = v }
 
     builder = ChangesetBuilder.new(@now, @conn)
@@ -207,6 +207,10 @@ class Replicator
     @config["data_dir"] + format("/%03d/%03d/%03d", sequence / 1000000, (sequence / 1000) % 1000, (sequence % 1000))
   end
 
+  def s3_stem
+    @config["s3_dir"] + format("/%03d/%03d/%03d", sequence / 1000000, (sequence / 1000) % 1000, (sequence % 1000))
+  end
+
   def write_tmp_files!(changesets)
     data_file = data_stem + ".osm.gz"
     tmp_state = @config["state_file"] + ".tmp"
@@ -216,9 +220,7 @@ class Replicator
     Zlib::GzipWriter.open(tmp_data) do |fh|
       fh.write(changeset_dump(changesets))
     end
-    File.open(tmp_state, "w") do |fh|
-      fh.write(YAML.dump(@state))
-    end
+    File.write(tmp_state, YAML.dump(@state))
 
     # fsync the files in their old locations.
     fsync(tmp_data)
@@ -257,12 +259,25 @@ class Replicator
     # from the dirent and has been updated to account for any allocations.
     fdirsync(File.dirname(data_file))
     fdirsync(File.dirname(@config["state_file"]))
+
+    if @config["s3_dir"]
+      s3_file = s3_stem + ".osm.gz"
+      s3_state_file = s3_stem + ".state.txt"
+
+      system("/opt/awscli/v2/current/bin/aws", "--profile=osm-pds-upload", "s3", "cp", "--storage-class=INTELLIGENT_TIERING", data_file, s3_file)
+      system("/opt/awscli/v2/current/bin/aws", "--profile=osm-pds-upload", "s3", "cp", "--storage-class=INTELLIGENT_TIERING", data_state_file, s3_state_file)
+    end
   end
 
   # saves new state (including the changeset dump xml)
   def save!
-    File.open(@config["state_file"], "r") do |fl|
-      fl.flock(File::LOCK_EX)
+    File.open(@config["lock_file"], File::RDWR | File::CREAT, 0o600) do |fl|
+      # take the lock in non-blocking mode. if this process doesn't get the lock
+      # then another will be run from cron shortly. this prevents a whole bunch
+      # of processes queueing on the lock and causing weirdness if/when they
+      # get woken up in a random order.
+      got_lock = fl.flock(File::LOCK_EX | File::LOCK_NB)
+      break unless got_lock
 
       # try and write the files to tmp locations and then
       # move them into place later, to avoid in-progress
@@ -277,8 +292,8 @@ class Replicator
         move_tmp_files_into_place!
 
         fl.flock(File::LOCK_UN)
-      rescue
-        STDERR.puts("Error! Couldn't update state.")
+      rescue StandardError
+        warn "Error! Couldn't update state."
         fl.flock(File::LOCK_UN)
         raise
       end
@@ -290,6 +305,9 @@ begin
   rep = Replicator.new(ARGV[0])
   rep.save!
 rescue StandardError => e
-  STDERR.puts "ERROR: #{e.message}"
+  warn "ERROR: #{e.message}"
+  e.backtrace.each do |frame|
+    warn "ERROR: #{frame}"
+  end
   exit 1
 end