Merge remote-tracking branch 'github/pull/89'

[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 8fd78a1bc33e1dfee2a81a535d26b24d30290d7e..e71d4c5482aae3a861f250de6ddde77accafc870 100644 (file)
--- a/cookbooks/planet/files/default/replication-bin/replicate-changesets
+++ b/cookbooks/planet/files/default/replication-bin/replicate-changesets
@@ -18,7 +18,7 @@ GEO_SCALE = 10000000
 ##
 # replace characters which cannot be represented in XML 1.0.
 def xml_sanitize(str)
-  str.gsub(/[\x00-\x08\x0b\x0c\x0e-\x20]/, "?")
+  str.gsub(/[\x00-\x08\x0b\x0c\x0e-\x1f]/, "?")
 end
 
 ##
@@ -117,6 +117,27 @@ class ChangesetBuilder
   end
 end
 
+##
+# sync a file to guarantee it's on disk
+def fsync(f)
+  File.open(f) do |fh|
+    fh.fsync
+  end
+end
+
+##
+# sync a directory to guarantee it's on disk. have to recurse to the root
+# to guarantee sync for newly created directories.
+def fdirsync(d)
+  while d != "/"
+    Dir.open(d) do |dh|
+      io = IO.for_fd(dh.fileno)
+      io.fsync
+    end
+    d = File.dirname(d)
+  end
+end
+
 ##
 # state and connections associated with getting changeset data
 # replicated to a file.
@@ -186,9 +207,11 @@ class Replicator
       fl.flock(File::LOCK_EX)
 
       sequence = (@state.key?("sequence") ? @state["sequence"] + 1 : 0)
-      data_file = @config["data_dir"] + format("/%03d/%03d/%03d.osm.gz", sequence / 1000000, (sequence / 1000) % 1000, (sequence % 1000))
+      data_stem = @config["data_dir"] + format("/%03d/%03d/%03d", sequence / 1000000, (sequence / 1000) % 1000, (sequence % 1000))
+      data_file = data_stem + ".osm.gz"
+      data_state_file = data_stem + ".state.txt"
       tmp_state = @config["state_file"] + ".tmp"
-      tmp_data = "/tmp/changeset_data.osm.tmp"
+      tmp_data = data_file + ".tmp"
       # try and write the files to tmp locations and then
       # move them into place later, to avoid in-progress
       # clashes, or people seeing incomplete files.
@@ -202,15 +225,37 @@ class Replicator
           fh.write(YAML.dump(@state))
         end
 
+        # fsync the files in their old locations.
+        fsync(tmp_data)
+        fsync(tmp_state)
+
+        # sync the directory as well, to ensure that the file is reachable
+        # from the dirent and has been updated to account for any allocations.
+        fdirsync(File.dirname(tmp_data))
+        fdirsync(File.dirname(tmp_state))
+
         # sanity check: the files we're moving into place
         # should be non-empty.
-        fail "Temporary gzip file should exist, but doesn't." unless File.exist?(tmp_data)
-        fail "Temporary state file should exist, but doesn't." unless File.exist?(tmp_state)
-        fail "Temporary gzip file should be non-empty, but isn't." if File.zero?(tmp_data)
-        fail "Temporary state file should be non-empty, but isn't." if File.zero?(tmp_state)
+        raise "Temporary gzip file should exist, but doesn't." unless File.exist?(tmp_data)
+        raise "Temporary state file should exist, but doesn't." unless File.exist?(tmp_state)
+        raise "Temporary gzip file should be non-empty, but isn't." if File.zero?(tmp_data)
+        raise "Temporary state file should be non-empty, but isn't." if File.zero?(tmp_state)
 
         FileUtils.mv(tmp_data, data_file)
-        FileUtils.mv(tmp_state, @config["state_file"])
+        FileUtils.cp(tmp_state, @config["state_file"])
+        FileUtils.mv(tmp_state, data_state_file)
+
+        # fsync the files in their new locations, in case the inodes have
+        # changed in the move / copy.
+        fsync(data_fie)
+        fsync(@config["state_file"])
+        fsync(data_state_file)
+
+        # sync the directory as well, to ensure that the file is reachable
+        # from the dirent and has been updated to account for any allocations.
+        fdirsync(File.dirname(data_file))
+        fdirsync(File.dirname(@config["state_file"]))
+
         fl.flock(File::LOCK_UN)
 
       rescue