README.md in fluent-plugin-buffered-metrics-0.0.1 vs README.md in fluent-plugin-buffered-metrics-0.0.2

@@ -55,11 +55,11 @@
     @type buffered_metrics
     metrics_backend graphite
     prefix fluentd.<hostname>.5m
     #url tcp://localhost:2003
     url file:///var/tmp/graphite_test.out
-    sum_defaults [{"-.-.-.count":0},{"-.-.-.bytes":0}]
+    sum_defaults [{"name":"-.-.-.count","value":0},{"name":"-.-.-.bytes","value":0}]
     sum_maps {"event['record'].empty? ? false : 1":["-.-.-.count","(['tag','facility','level'].map {|t| event['record'][t] || '-'}+['count']).join('.')"],"event['record'].empty? ? false : event.to_s.length":["-.-.-.bytes","(['tag','facility','level'].map {|t| event['record'][t] || '-'}+['bytes']).join('.')"]}
     metric_maps {}
     metric_defaults []
     flush_interval 5m
     # Make this a working configuration
@@ -104,14 +104,16 @@
 
 #### sum_defaults:
 ```JSON
   [
      {
-        "-.-.-.count" : 0
+        "name": "-.-.-.count",
+        "value": 0
      },
      {
-        "-.-.-.bytes" : 0
+        "name": "-.-.-.bytes",
+        "value": 0
      }
   ]
 ```
 
-The "defaults" specifications are needed if there are metrics which need to be sent with every metrics publication, even it none of the matches are met and the metric is never created during the run.  This may be more of an artifact, as this was initially done for Stackdriver, which had real issues with gaps in metrics series.
\ No newline at end of file
+The "defaults" specifications are needed if there are metrics which need to be sent with every metrics publication, even it none of the matches are met and the metric is never created during the run.  This may be more of an artifact, as this was initially done for Stackdriver, which had real issues with gaps in metrics series.