[Telemetry] Refactor common Android cleanup functions.

build/android/pylib/utils/test_environment.py

+# Copyright (c) 2013 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 os
+import psutil
+
+from pylib import android_commands
+from pylib import ports
+
+def _KillWebServers():
+  for retry in range(5):

[tonyg, 2013/11/25 15:46:33]

Style nit: use xrange when it is constant.

[bulach, 2013/11/27 12:10:53]

just in case :) python3 has removed xrange, there's only range there..
not that we'd change anytime soon, but anyways... :)

[Primiano Tucci (use gerrit), 2013/11/27 15:14:41]

Ah right (both). It looks, however, that there is enough xrange precedent in
chromium, so someone at some point will have to remove the x when we'll move to
Py3.
On the other side we won't spend 700 nanoseconds just for a missing 'x'
(http://pastebin.com/j7Pa17vL, I did some quick measurements)

+    for server in ['lighttpd', 'web-page-replay']:
+      pids = [p.pid for p in psutil.process_iter() if server in p.name]
+      for pid in pids:
+        try:
+          logging.warning('Killing %s %s', server, pid)
+          os.kill(pid, signal.SIGQUIT)
+        except Exception as e:
+          logging.warning('Failed killing %s %s %s', server, pid, e)
+
+
+def CleanupPendingProcesses():

[tonyg, 2013/11/25 15:46:33]

Would CleanupLeftoverProcesses() be a better name?

[Primiano Tucci (use gerrit), 2013/11/25 17:26:51]

Why not, sounds good to me.

+  """Clean up the test environment, restarting fresh adb and HTTP daemons."""
+  _KillWebServers()
+  did_restart_host_adb = False
+  for device in android_commands.GetAttachedDevices():

[tonyg, 2013/11/25 15:46:33]

Is this a problem for sharding? Could we kill an adb that is in use by another
shard?

[Primiano Tucci (use gerrit), 2013/11/25 17:26:51]

Hmm, to be honest I have no idea how sharding is  handled. I wrote this piece of
code following Marcus' suggestions. Therefore, I invoke the supreme help of the
sharding master to answer this question! :)

[bulach, 2013/11/27 12:10:53]

tl;dr; it's not a problem as it is right now :)

long version:
- for bots, build/android/pylib/perf/setup.py kicks off at the "shard master"
process, before any actual "shard slave" has started, and calls this.. nothing
else should call this method.

- for developers, only
tools/telemetry/telemetry/core/backends/chrome/android_browser_finder.py calls
this... it's guarded there so when running in buildbot, the shards won't mess up
with each other..

it's a bit brittle since it relies on the caller, but anyways...

+    adb = android_commands.AndroidCommands(device)
+    # Make sure we restart the host adb server only once.
+    if not did_restart_host_adb:
+      adb.RestartAdbServer()
+      did_restart_host_adb = True
+    adb.RestartAdbdOnDevice()
+    adb.EnableAdbRoot()
+    adb.WaitForDevicePm()