Avoid forwarding the test HTTP port twice.

Avoid forwarding the test HTTP port twice.

base_test_runner.BaseTestRunner already forwards port 8000; avoid doing that
again in instrumentation.test_runner.TestRunner again, which confuses the
forwarder.

BUG=242846
Committed: https://src.chromium.org/viewvc/chrome?view=rev&revision=210666

[Joao da Silva, 2013-07-09 14:39:26]

https://codereview.chromium.org/18854020/diff/1/build/android/pylib/instrumen...
File build/android/pylib/instrumentation/test_runner.py (right):

https://codereview.chromium.org/18854020/diff/1/build/android/pylib/instrumen...
build/android/pylib/instrumentation/test_runner.py:149: http_server_ports =
self.LaunchTestHttpServer(
Basically, this line already starts forwarding for the http_server_ports, so
doing it again below is redundant and was confusing the forwarder.

[pliard, 2013-07-09 14:42:08]

LGTM, Thanks a lot Joao! As I told you offline I have just landed r210557 to
workaround the issue in forwarder itself but fixing the root cause is great!

[Joao da Silva, 2013-07-09 15:19:03]

So here's what I've found:

- without this patch and without https://codereview.chromium.org/18051018 then
the pythonDrivenTests always fail, because they can't connect to localhost:8000

- with either patch then the test passes once after manually killing the server;
subsequent runs fail again

This CL may be considered a cleanup and I leave it you to determine if it's
worth landing it now or not.

We still need a way of ensuring that the server is sane before running a test
(e.g. by killing it).

Philippe, WDYT?

[pliard, 2013-07-09 15:28:00]

On 2013/07/09 15:19:03, Joao da Silva wrote:
> So here's what I've found:
> 
> - without this patch and without https://codereview.chromium.org/18051018 then
> the pythonDrivenTests always fail, because they can't connect to
localhost:8000
> 
> - with either patch then the test passes once after manually killing the
server;
> subsequent runs fail again
> 
> This CL may be considered a cleanup and I leave it you to determine if it's
> worth landing it now or not.
> 
> We still need a way of ensuring that the server is sane before running a test
> (e.g. by killing it).
> 
> Philippe, WDYT?

I think this CL is still valuable thus should still land. I'm preparing a CL
modifying python_test_sharder.py to kill the forwarder before the tests run.
This action is normally performed in build/android/pylib/base/shard.py but the
Python-driven tests don't seem to reuse that code.

I'm also working on a CL in parallel to address the underlying issue in
forwarder itself.

Thanks a lot Joao!

[Joao da Silva, 2013-07-09 15:47:52]

Marcus, can you do an owner check? Thanks!

[bulach, 2013-07-09 18:15:02]

lgtm, thanks!

[commit-bot: I haz the power, 2013-07-09 18:43:15]

CQ is trying da patch. Follow status at
https://chromium-status.appspot.com/cq/joaodasilva@chromium.org/18854020/1

[commit-bot: I haz the power, 2013-07-09 23:42:16]

Change committed as 210666