Thanks for the report. I think the assertion is copied over from the other tests and is not crucial for this test case. I will get rid of it.

No problem.

I did notice there are some other instances where this test also just did not complete. Perhaps it's too long-running on Dr. Memory bots.

The following revision refers to this bug:
  https://chromium.googlesource.com/chromium/src.git/+/fe811567d227f4e509da5417b459549ad6c3c807

commit fe811567d227f4e509da5417b459549ad6c3c807
Author: tommycli <tommycli@chromium.org>
Date: Thu May 12 21:44:48 2016

Memory Sheriff: Disable HttpStreamFactoryBidirectionalQuicTest.RequestBidirectionalStreamImplQuicAlternative on Dr. Memory.

BUG= 611533 
TBR=

Review-Url: https://codereview.chromium.org/1976693003
Cr-Commit-Position: refs/heads/master@{#393364}

[modify] https://crrev.com/fe811567d227f4e509da5417b459549ad6c3c807/tools/valgrind/gtest_exclude/net_unittests.gtest-drmemory_win32.txt

https://build.chromium.org/p/chromium.memory.fyi/builders/Windows%20Unit%20%28DrMemory%29/builds/4842/steps/memory%20test%3A%20net/logs/stdio also gave:

[ RUN      ] Version/HttpStreamFactoryBidirectionalQuicTest.RequestBidirectionalStreamImplHttpJobFailsQuicJobSucceeds/0
c:\b\build\slave\drm-cr\build\src\net\http\http_stream_factory_impl_unittest.cc(1773): error: Value of: GetSocketPoolGroupCount(session()->GetTransportSocketPool( HttpNetworkSession::NORMAL_SOCKET_POOL))
  Actual: 1
Expected: 0
c:\b\build\slave\drm-cr\build\src\net\http\http_stream_factory_impl_unittest.cc(1775): error: Value of: GetSocketPoolGroupCount(session()->GetSSLSocketPool( HttpNetworkSession::NORMAL_SOCKET_POOL))
  Actual: 1
Expected: 0

The following revision refers to this bug:
  https://chromium.googlesource.com/chromium/src.git/+/e6c0324b141332dd7366e2964c0ed58fe171352d

commit e6c0324b141332dd7366e2964c0ed58fe171352d
Author: jyasskin <jyasskin@chromium.org>
Date: Sat May 14 01:04:00 2016

HttpStreamFactoryBidirectionalQuicTest.RequestBidirectionalStreamImplHttpJobFailsQuicJobSucceeds is also flaky.

BUG= 611533 
TBR=xunjieli@chromium.org

Review-Url: https://codereview.chromium.org/1977053003
Cr-Commit-Position: refs/heads/master@{#393717}

[modify] https://crrev.com/e6c0324b141332dd7366e2964c0ed58fe171352d/tools/valgrind/gtest_exclude/net_unittests.gtest-drmemory_win32.txt

These two unit tests (RequestBidirectionalStreamImplHttpJobFailsQuicJobSucceeds and RequestBidirectionalStreamImplQuicAlternative) are flaky on DrMemory bots. We do not understand why the TCP sockets are established in these test runs (see discussion with rch@ in https://codereview.chromium.org/1977513003/), so we decided to leave these assertions alone.