Try to wait for devices to come back after a perf test failure.

Try to wait for devices to come back after a perf test failure.

I notice a common pattern on the perf bots where on test fails, then
all subsequent tests on that shard fail, but the device is magically
back on the next run. So I suspect that the device is recovering
itself, perhaps after a reboot.

BUG=

Committed: https://crrev.com/aafdf82b4605befff8288ab5a0bfc5eccd762bae
Cr-Commit-Position: refs/heads/master@{#299743}

[tonyg, 2014-10-14 00:55:13]

tonyg@chromium.org changed reviewers:
+ jbudorick@chromium.org

[jbudorick, 2014-10-15 08:06:18]

Sorry for the delay.

https://codereview.chromium.org/649423002/diff/1/build/android/pylib/perf/tes...
File build/android/pylib/perf/test_runner.py (right):

https://codereview.chromium.org/649423002/diff/1/build/android/pylib/perf/tes...
build/android/pylib/perf/test_runner.py:65: sys.path.append(os.path.join(
(see comment below first)

this would become

  from pylib.device import device_errors

https://codereview.chromium.org/649423002/diff/1/build/android/pylib/perf/tes...
build/android/pylib/perf/test_runner.py:267:
self.device.old_interface.WaitForDevicePm()
I'd rather see this use:

  self.device.WaitUntilFullyBooted()

as your comment above indicates that that's kind of what you're trying to do,
and we're trying to avoid adding any new direct references to AndroidCommands
via old_interface.

https://codereview.chromium.org/649423002/diff/1/build/android/pylib/perf/tes...
build/android/pylib/perf/test_runner.py:268: except
errors.WaitForResponseTimedOutError as e:
this would become

  except device_errors.CommandTimeoutError as e:

[tonyg, 2014-10-15 16:47:27]

Oops, sorry for trying to use the deprecated version and thanks for pointing me
towards the light :)

https://codereview.chromium.org/649423002/diff/1/build/android/pylib/perf/tes...
File build/android/pylib/perf/test_runner.py (right):

https://codereview.chromium.org/649423002/diff/1/build/android/pylib/perf/tes...
build/android/pylib/perf/test_runner.py:65: sys.path.append(os.path.join(
On 2014/10/15 08:06:18, jbudorick wrote:
> (see comment below first)
> 
> this would become
> 
>   from pylib.device import device_errors

Done.

https://codereview.chromium.org/649423002/diff/1/build/android/pylib/perf/tes...
build/android/pylib/perf/test_runner.py:267:
self.device.old_interface.WaitForDevicePm()
On 2014/10/15 08:06:18, jbudorick wrote:
> I'd rather see this use:
> 
>   self.device.WaitUntilFullyBooted()
> 
> as your comment above indicates that that's kind of what you're trying to do,
> and we're trying to avoid adding any new direct references to AndroidCommands
> via old_interface.

Done.

https://codereview.chromium.org/649423002/diff/1/build/android/pylib/perf/tes...
build/android/pylib/perf/test_runner.py:268: except
errors.WaitForResponseTimedOutError as e:
On 2014/10/15 08:06:18, jbudorick wrote:
> this would become
> 
>   except device_errors.CommandTimeoutError as e:

Done.

[jbudorick, 2014-10-15 17:33:25]

lgtm

[tonyg, 2014-10-15 18:22:36]

The CQ bit was checked by tonyg@chromium.org

[commit-bot: I haz the power, 2014-10-15 18:23:27]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/649423002/20001

[commit-bot: I haz the power, 2014-10-15 19:32:56]

Committed patchset #2 (id:20001)

[commit-bot: I haz the power, 2014-10-15 19:33:44]

Patchset 2 (id:??) landed as
https://crrev.com/aafdf82b4605befff8288ab5a0bfc5eccd762bae
Cr-Commit-Position: refs/heads/master@{#299743}