Fix off-by-one issue when generating statistics

CHANGELOG.md

@@ -1,6 +1,7 @@
 ## 0.2.0 (unreleased)
 
   - Add a TCP "aliveness_check" option for easier monitoring.
+  - Fix off-by-one issue in calculating percentile statistics.
 
 ## 0.1.0 (August 16, 2011)
 

statsite/metrics.py

@@ -93,11 +93,19 @@ def fold(cls, lst, now, percentile=90):
             lower_idx = (len(vals) - inner_indexes) / 2
             upper_idx = lower_idx + inner_indexes
 
-            val_sum_pct = sum(vals[lower_idx:upper_idx+1])
+            # If we only have one item, then the percentile is just the
+            # values itself, otherwise the lower_idx:upper_idx slice returns
+            # an empty list.
+            if len(vals) == 1:
+                vals_pct = vals
+            else:
+                vals_pct    = vals[lower_idx:upper_idx]
+
+            val_sum_pct = sum(vals_pct)
             val_avg_pct = val_sum_pct / inner_indexes if inner_indexes > 0 else val_sum_pct
             val_min_pct = vals[lower_idx]
             val_max_pct = vals[upper_idx]
-            val_stdev_pct = cls._stdev(vals[lower_idx:upper_idx+1], val_avg_pct)
+            val_stdev_pct = cls._stdev(vals_pct, val_avg_pct)
 
             outputs.append(("timers.%s.sum" % key, val_sum, now))
             outputs.append(("timers.%s.mean" % key, val_avg, now))

tests/unit/test_metric_timer.py

@@ -90,6 +90,22 @@ def test_fold_stdev(self):
         assert ("timers.k.stdev", Timer._stdev([10, 15], 12.5), now) == self._get_metric("timers.k.stdev", result)
         assert ("timers.j.stdev", Timer._stdev([7.9, 8], 7.95), now) == self._get_metric("timers.j.stdev", result)
 
+    def test_sum_percentile(self):
+        """
+        Tests the percentile sum is properly counted.
+        """
+        now = 10
+        result = Timer.fold(self._100_timers, now)
+        assert ("timers.k.sum_90", 4545, now) == self._get_metric("timers.k.sum_90", result)
+
+    def test_mean_percentile(self):
+        """
+        Tests the percentile sum is properly counted.
+        """
+        now = 10
+        result = Timer.fold(self._100_timers, now)
+        assert ("timers.k.mean_90", 50, now) == self._get_metric("timers.k.mean_90", result)
+
     def test_stdev(self):
         """
         Tests that the standard deviation is properly computed.
@@ -109,3 +125,11 @@ def _get_metric(self, key, metrics):
                 return metric
 
         return None
+
+    @property
+    def _100_timers(self):
+        result = []
+        for i in xrange(1, 101):
+            result.append(Timer("k", i))
+
+        return result