QUIC - Disable QUIC when there is bad packet loss rate (50% of the

QUIC - Disable QUIC when there is bad packet loss rate (50% of the
packets are lost) for 4 consecutive crypto handshakes.

For a session, during crypto handshake if there is a bad packet loss
rate, then we mark that QUIC is recently broken for that host/port.

We disable QUIC and close that connections if 4 consecutive handshakes
have bad packet loss rate. QUIC was disabled for 0.002% of the time.

We have been running the above experiment in M45 (in Canary, Dev and
Beta). In M45 Beta, Net.HttpJob.TotalTimeNotCached.Quic's mean is
faster by 4.4% (for data aggreagted for 28 days) when this experiment
was enabled.

R=rch@chromium.org

Committed: https://crrev.com/147024676acbf52d08179457d2909314cf423a5c
Cr-Commit-Position: refs/heads/master@{#347871}

[Ryan Hamilton, 2015-09-08 17:53:41]

lgtm

https://codereview.chromium.org/1326763007/diff/1/net/http/http_network_sessi...
File net/http/http_network_session.cc (right):

https://codereview.chromium.org/1326763007/diff/1/net/http/http_network_sessi...
net/http/http_network_session.cc:107: quic_packet_loss_threshold(0.5f),
Do you need to change io_thread_unittests.cc to reflect the new defaults?

[ramant (doing other things), 2015-09-08 19:46:10]

Thanks much Ryan. Fixed the io_thread_unittests.

https://codereview.chromium.org/1326763007/diff/1/net/http/http_network_sessi...
File net/http/http_network_session.cc (right):

https://codereview.chromium.org/1326763007/diff/1/net/http/http_network_sessi...
net/http/http_network_session.cc:107: quic_packet_loss_threshold(0.5f),
On 2015/09/08 17:53:41, Ryan Hamilton wrote:
> Do you need to change io_thread_unittests.cc to reflect the new defaults?

Done.

[ramant (doing other things), 2015-09-08 21:34:01]

Patchset #2 (id:20001) has been deleted

[ramant (doing other things), 2015-09-08 21:34:38]

The CQ bit was checked by rtenneti@chromium.org

[ramant (doing other things), 2015-09-08 21:34:39]

The patchset sent to the CQ was uploaded after l-g-t-m from rch@chromium.org
Link to the patchset: https://codereview.chromium.org/1326763007/#ps40001
(title: "Fix unit tests and merge with TOT")

[commit-bot: I haz the power, 2015-09-08 21:34:52]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/1326763007/40001
View timeline at
 https://chromium-cq-status.appspot.com/patch-timeline/1326763007/40001

[commit-bot: I haz the power, 2015-09-08 23:00:07]

The CQ bit was unchecked by commit-bot@chromium.org

[commit-bot: I haz the power, 2015-09-08 23:00:12]

Try jobs failed on following builders:
  linux_android_rel_ng on tryserver.chromium.linux (JOB_FAILED,
http://build.chromium.org/p/tryserver.chromium.linux/builders/linux_android_r...)

[ramant (doing other things), 2015-09-09 03:02:45]

The CQ bit was checked by rtenneti@chromium.org

[commit-bot: I haz the power, 2015-09-09 03:03:20]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/1326763007/40001
View timeline at
 https://chromium-cq-status.appspot.com/patch-timeline/1326763007/40001

[commit-bot: I haz the power, 2015-09-09 09:07:34]

Committed patchset #2 (id:40001)

[commit-bot: I haz the power, 2015-09-09 09:08:43]

Patchset 2 (id:??) landed as
https://crrev.com/147024676acbf52d08179457d2909314cf423a5c
Cr-Commit-Position: refs/heads/master@{#347871}

[ramant (doing other things), 2015-10-21 18:17:19]

A revert of this CL (patchset #2 id:40001) has been created in
https://codereview.chromium.org/1409143007/ by rtenneti@chromium.org.

The reason for reverting is: BUG=546089

We want to run bad packet loss as an experiment because on youtube, it had a
negative impact..