Restart servers before retrying layout tests

Restart servers before retrying layout tests

crbug.com/720511 seems caused by some server issues. The flaky tests
often fail when run with all other tests but almost always succeed
if run separately. I guess some test run before these tests caused
the server in a special state to return "internal error" for the tests.

Restarting the server might let the tests pass on retry to avoid the
tests from breaking bots.

BUG=720511

[Xianzhu, 2017-05-11 17:21:19]

wangxianzhu@chromium.org changed reviewers:
+ qyearsley@chromium.org, wkorman@chromium.org

[Xianzhu, 2017-05-11 17:21:45]

Description was changed from

==========
Restart servers before retrying layout tests

crbug.com/720511 seems caused by some server issues. The flaky tests
often fail when run with all other tests but almost always succeed
if run separately. I guess some test run before these tests caused
the server in a special state to return "internal error" for the tests.

Restart the server might let the tests pass on retry.

BUG=720511
==========

to

==========
Restart servers before retrying layout tests

crbug.com/720511 seems caused by some server issues. The flaky tests
often fail when run with all other tests but almost always succeed
if run separately. I guess some test run before these tests caused
the server in a special state to return "internal error" for the tests.

Restarting the server might let the tests pass on retry to avoid the
tests from breaking bots.

BUG=720511
==========

[Xianzhu, 2017-05-11 17:28:27]

The CQ bit was checked by wangxianzhu@chromium.org to run a CQ dry run

[commit-bot: I haz the power, 2017-05-11 17:29:20]

Dry run: CQ is trying da patch.

Follow status at:
https://chromium-cq-status.appspot.com/v2/patch-status/codereview.chromium.or...

[Xianzhu, 2017-05-11 18:09:06]

Unfortunately, this doesn't fix the issue. The tests still failed in retries.

[qyearsley, 2017-05-11 20:07:46]

On 2017/05/11 at 18:09:06, wangxianzhu wrote:
> Unfortunately, this doesn't fix the issue. The tests still failed in retries.

So this CL was a test for a potential fix but it appears that this doesn't fix
the issue so we don't want to commit this, right?

[Xianzhu, 2017-05-11 20:44:05]

On 2017/05/11 20:07:46, qyearsley wrote:
> On 2017/05/11 at 18:09:06, wangxianzhu wrote:
> > Unfortunately, this doesn't fix the issue. The tests still failed in
retries.
> 
> So this CL was a test for a potential fix but it appears that this doesn't fix
> the issue so we don't want to commit this, right?

Yes.

I was quite confident when requesting review, but turned out to be misled by
crbug.com/721466.