X-Git-Url: https://git.openstreetmap.org/chef.git/blobdiff_plain/276750ebe30538b39c2d756dc4e784afa2c974ab..18e12b0e123fc3e5f344682f884feabc12da0f5a:/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 old mode 100644 new mode 100755 index b5e6b50d9..fac548bd5 --- a/cookbooks/planet/files/default/replication-bin/replicate-changesets +++ b/cookbooks/planet/files/default/replication-bin/replicate-changesets @@ -100,7 +100,7 @@ class ChangesetBuilder xml["comments_count"] = res.num_tuples.to_s # early return if there aren't any comments - return unless res.num_tuples > 0 + return unless res.num_tuples.positive? discussion = XML::Node.new("discussion") res.each do |row| @@ -141,16 +141,16 @@ end # replicated to a file. class Replicator def initialize(config) - @config = YAML.load(File.read(config)) - @state = YAML.load(File.read(@config["state_file"])) + @config = YAML.safe_load(File.read(config)) + @state = YAML.safe_load(File.read(@config["state_file"]), [Time]) @conn = PGconn.connect(@config["db"]) - @now = Time.now.getutc + # 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 def open_changesets last_run = @state["last_run"] last_run = (@now - 60) if last_run.nil? - @state["last_run"] = @now # pretty much all operations on a changeset will modify its closed_at # time (see rails_port's changeset model). so it is probably enough # for us to look at anything that was closed recently, and filter from @@ -167,7 +167,7 @@ class Replicator new_ids = @conn .exec("select distinct changeset_id from changeset_comments where created_at >= '#{last_run}' and created_at < '#{@now}' and visible") .map { |row| row["changeset_id"].to_i } - .select { |c_id| !cs_ids.include?(c_id) } + .reject { |c_id| cs_ids.include?(c_id) } new_ids.each do |id| @conn @@ -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) @@ -199,19 +199,22 @@ class Replicator doc.to_s end + def sequence + @state.key?("sequence") ? @state["sequence"] + 1 : 0 + end + def data_stem - sequence = (@state.key?("sequence") ? @state["sequence"] + 1 : 0) @config["data_dir"] + format("/%03d/%03d/%03d", sequence / 1000000, (sequence / 1000) % 1000, (sequence % 1000)) end - def write_tmp_files! + def write_tmp_files!(changesets) data_file = data_stem + ".osm.gz" tmp_state = @config["state_file"] + ".tmp" tmp_data = data_file + ".tmp" FileUtils.mkdir_p(File.dirname(data_file)) Zlib::GzipWriter.open(tmp_data) do |fh| - fh.write(changeset_dump(open_changesets)) + fh.write(changeset_dump(changesets)) end File.open(tmp_state, "w") do |fh| fh.write(YAML.dump(@state)) @@ -246,7 +249,7 @@ class Replicator # fsync the files in their new locations, in case the inodes have # changed in the move / copy. - fsync(data_fie) + fsync(data_file) fsync(@config["state_file"]) fsync(data_state_file) @@ -265,15 +268,16 @@ class Replicator # move them into place later, to avoid in-progress # clashes, or people seeing incomplete files. begin + changesets = open_changesets @state["sequence"] = sequence + @state["last_run"] = @now - write_tmp_files! + write_tmp_files!(changesets) move_tmp_files_into_place! fl.flock(File::LOCK_UN) - - rescue + rescue StandardError STDERR.puts("Error! Couldn't update state.") fl.flock(File::LOCK_UN) raise