[draft] Modifying telemetry to collect memory dumps over a large set of sites

tools/perf/page_sets/desktop_memory.py

Index: tools/perf/page_sets/desktop_memory.py
diff --git a/tools/perf/page_sets/desktop_memory.py b/tools/perf/page_sets/desktop_memory.py
new file mode 100644
index 0000000000000000000000000000000000000000..7f6d781931ab7edbcbe6de6524a81f1a800c9dda
--- /dev/null
+++ b/tools/perf/page_sets/desktop_memory.py
@@ -0,0 +1,75 @@
+# 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 py_utils
+
+from telemetry.page import page as page_module
+from telemetry.page import shared_page_state
+from telemetry import story
+
+_DUMP_WAIT_TIME = 3
+_ITERATIONS = 10
+
+class DesktopMemoryPage(page_module.Page):
+
+  def __init__(self, url, page_set):
+    super(DesktopMemoryPage, self).__init__(
+        url=url, page_set=page_set,
+        shared_page_state_class=shared_page_state.SharedDesktopPageState)
+
+  def _DumpMemory(self, action_runner, phase):
+    with action_runner.CreateInteraction(phase):
+      action_runner.Wait(_DUMP_WAIT_TIME)
+      action_runner.ForceGarbageCollection()
+      action_runner.SimulateMemoryPressureNotification('critical')
+      action_runner.Wait(_DUMP_WAIT_TIME)
+      # action_runner.MeasureMemory(deterministic_mode=False)
+      action_runner.tab.browser.DumpMemory()
+
+  def RunPageInteractions(self, action_runner):
+    self._DumpMemory(action_runner, 'pre')
+    for _ in xrange(_ITERATIONS):
+      action_runner.ReloadPage()
+
+    tabs = action_runner.tab.browser.tabs
+    for _ in xrange(_ITERATIONS):
+      new_tab = tabs.New()
+      new_tab.action_runner.Navigate(self._url)
+      try:
+        new_tab.action_runner.WaitForNetworkQuiescence()
+      except py_utils.TimeoutException:
+        logging.warning('WaitForNetworkQuiescence() timeout')
+      new_tab.Close()
+
+    self._DumpMemory(action_runner, 'post')
+
+
+class DesktopMemoryPageSet(story.StorySet):
+
+  """ Desktop sites with interesting memory characteristics """
+
+  def __init__(self):
+    super(DesktopMemoryPageSet, self).__init__(
+        archive_data_file='data/mobile_memory.json',
+        cloud_storage_bucket=story.PARTNER_BUCKET)
+
+    urls_list = [
+      'http://www.perdu.com',
+      'http://www.google.com',
+      "http://www.live.com",
+      "http://www.youtube.com",
+      "http://www.wikipedia.org",
+      "http://www.flickr.com/",
+      "http://www.cnn.com/",
+      "http://www.adobe.com/",
+      "http://www.aol.com/",
+      "http://www.cnet.com/",
+      "http://www.godaddy.com/",
+      "http://www.walmart.com/",
+      "http://www.skype.com/",
+    ]
+
+    for url in urls_list:
+      self.AddStory(DesktopMemoryPage(url, self))
+