Add mitigate code for existing code

Signed-off-by: Yuta Iwama <[email protected]>
fluent · Feb 14, 2020 · 0538d26 · 0538d26
1 parent 71a346d
commit 0538d26
Show file tree

Hide file tree

Showing 2 changed files with 23 additions and 4 deletions.
diff --git a/lib/fluent/plugin/buf_file.rb b/lib/fluent/plugin/buf_file.rb
@@ -166,7 +166,16 @@ def resume
 
           case chunk.state
           when :staged
-            stage[chunk.metadata] = chunk
+            # unstaged chunk created at Buffer#write_step_by_step is identified as the staged chunk here because FileChunk#assume_chunk_state checks only the file name.
+            # https://github.com/fluent/fluentd/blob/9d113029d4550ce576d8825bfa9612aa3e55bff0/lib/fluent/plugin/buffer.rb#L663
+            # This case can happen when fluentd process is killed by signal or other reasons between creating unstaged chunks and changing them to staged mode in Buffer#write
+            # these chunks(unstaged chunks) has shared the same metadata
+            # So perform enqueue step again https://github.com/fluent/fluentd/blob/9d113029d4550ce576d8825bfa9612aa3e55bff0/lib/fluent/plugin/buffer.rb#L364
+            if chunk_size_full?(chunk) || stage.key?(chunk.metadata)
+              queue << chunk.enqueued!
+            else
+              stage[chunk.metadata] = chunk
+            end
           when :queued
             queue << chunk
           end
@@ -181,7 +190,7 @@ def generate_chunk(metadata)
         # FileChunk generates real path with unique_id
         perm = @file_permission || system_config.file_permission
         chunk = Fluent::Plugin::Buffer::FileChunk.new(metadata, @path, :create, perm: perm, compress: @compress)
-        log.debug "Created new chunk", chunk_id: dump_unique_id_hex(chunk.unique_id), metadata: metadata
+        log.info "Created new chunk", chunk_id: dump_unique_id_hex(chunk.unique_id), metadata: metadata
 
         return chunk
       end

diff --git a/test/plugin/test_buf_file.rb b/test/plugin/test_buf_file.rb
@@ -1032,7 +1032,15 @@ def write_metadata(path, chunk_id, metadata, size, ctime, mtime)
         f.write ["t4.test", event_time('2016-04-17 14:00:28 -0700').to_i, {"message" => "yay6"}].to_json + "\n"
       end
       m3 = m2.dup_next
-      write_metadata(p3 + '.meta', c1id, m3, 2, event_time('2016-04-17 14:00:00 -0700').to_i, event_time('2016-04-17 14:00:28 -0700').to_i)
+      write_metadata(p3 + '.meta', c3id, m3, 2, event_time('2016-04-17 14:00:00 -0700').to_i, event_time('2016-04-17 14:00:28 -0700').to_i)
+
+      c4id = Fluent::UniqueId.generate
+      p4 = File.join(@bufdir, "testbuf.b#{Fluent::UniqueId.hex(c4id)}.log")
+      File.open(p4, 'wb') do |f|
+        f.write ["t1.test", event_time('2016-04-17 14:00:15 -0700').to_i, {"message" => "yay5"}].to_json + "\n"
+        f.write ["t4.test", event_time('2016-04-17 14:00:28 -0700').to_i, {"message" => "yay6"}].to_json + "\n"
+      end
+      write_metadata(p4 + '.meta', c4id, m3, 2, event_time('2016-04-17 14:00:00 -0700').to_i, event_time('2016-04-17 14:00:28 -0700').to_i)
 
       Fluent::Test.setup
       @d = FluentPluginFileBufferTest::DummyOutputPlugin.new
@@ -1060,9 +1068,11 @@ def write_metadata(path, chunk_id, metadata, size, ctime, mtime)
     end
 
     test '#resume returns each chunks' do
-      s, _ = @p.resume
+      s, e = @p.resume
       assert_equal 3, s.size
       assert_equal [0, 1, 2], s.keys.map(&:seq).sort
+      assert_equal 1, e.size
+      assert_equal [2], e.map { |e| e.metadata.seq }
     end
   end