Add new Telemetry tests to get CPU and memory usage for idle and flinging two test scenarios. Mirro…

chrome/test/media_router/telemetry/benchmarks/media_router_cpu_memory_metric.py

Index: chrome/test/media_router/telemetry/benchmarks/media_router_cpu_memory_metric.py
diff --git a/chrome/test/media_router/telemetry/benchmarks/media_router_cpu_memory_metric.py b/chrome/test/media_router/telemetry/benchmarks/media_router_cpu_memory_metric.py
new file mode 100644
index 0000000000000000000000000000000000000000..779dab18477b9d7606aafa0c4a299b8662f08f4d
--- /dev/null
+++ b/chrome/test/media_router/telemetry/benchmarks/media_router_cpu_memory_metric.py
@@ -0,0 +1,47 @@
+# Copyright 2016 The Chromium Authors. All rights reserved.
+# Use of this source code is governed by a BSD-style license that can be
+# found in the LICENSE file.
+
+import logging
+import json
+
+from telemetry.core import exceptions
+from telemetry.value import scalar
+
+from metrics import Metric
+
+
+class MediaRouterCPUMemoryMetric(Metric):
+  "A metric for media router CPU/Memory usage."
+
+  def Start(self, page, tab):
+    raise NotImplementedError()
+
+  def Stop(self, page, tab):
+    raise NotImplementedError()
+
+  def AddResults(self, tab, results):
+    results_json = None
+    try:
+      results_json = tab.EvaluateJavaScript(
+          'JSON.stringify(window.perfResults)')
+    except exceptions.EvaluateException:
+      pass
+    # This log gives the detailed information about CPU/memory usage.
+    print 'results_json' + ': ' + str(results_json)

[mark a. foltz, 2016/04/04 20:25:06]

Is this for debugging?  If so put it behind e.g. logging.info() so logging
doesn't affect normal test performance.

[Lei Lei, 2016/04/13 00:22:43]

Done.

+
+    if results_json:
+      perf_results = json.loads(results_json)
+      for metric in perf_results:
+        units = ''

[mark a. foltz, 2016/04/04 20:25:06]

What are the units for metrics other than privateMemory?

[Lei Lei, 2016/04/13 00:22:43]

Right now we only have two metrics, privateMemory and CPU. The units for
privateMemory is MB, the units for CPU is empty string.

I defined a constant for the units of each metric, which makes the code easier
to read.

+        metric_results = perf_results[metric]
+        for process in metric_results:

[mark a. foltz, 2016/04/04 20:25:06]

Is metric_results a dictionary? If so, I think this can be written

for (process, process_results) in perf_results[metric]

[Lei Lei, 2016/04/13 00:22:43]

Done.

+          process_results = metric_results[process]
+          if len(process_results) > 0:
+            avg_result = round(sum(process_results)/len(process_results), 4)
+            if metric == 'privateMemory':
+              avg_result = round(avg_result/(1024 * 1024), 2)
+              units = 'MB'

[mark a. foltz, 2016/04/04 20:25:06]

Do you need to reset this if metric != privateMemory?

[Lei Lei, 2016/04/13 00:22:43]

Yes, the units is reset at the beginning of for loop, see line 36.

+            results.AddValue(scalar.ScalarValue(
+                results.current_page, '%s_%s' % (process, metric), units,
+                avg_result))