All graphs for this bug:
  https://chromeperf.appspot.com/group_report?bug_id=607838

Original alerts at time of bug-filing:
  https://chromeperf.appspot.com/group_report?keys=agxzfmNocm9tZXBlcmZyFAsSB0Fub21hbHkYgICg5MX3pwoM,agxzfmNocm9tZXBlcmZyFAsSB0Fub21hbHkYgICg5MKJvgkM,agxzfmNocm9tZXBlcmZyFAsSB0Fub21hbHkYgICg5P2fowoM,agxzfmNocm9tZXBlcmZyFAsSB0Fub21hbHkYgICglNDBrgsM,agxzfmNocm9tZXBlcmZyFAsSB0Fub21hbHkYgICg5KXotgoM,agxzfmNocm9tZXBlcmZyFAsSB0Fub21hbHkYgICg5Lu3qwoM,agxzfmNocm9tZXBlcmZyFAsSB0Fub21hbHkYgICg5MPMoAoM


Bot(s) for this bug's original alert(s):

webrtc-android-tests-nexus5

@pbos: It seems that the CL https://codereview.webrtc.org/1917083003 is the most likely cause for this. Could you please take a look at this?

 Issue 607836  has been merged into this issue.

The following revision refers to this bug:
  https://chromium.googlesource.com/external/webrtc.git/+/602316c3cd8556cc78d44f3ea4cd5fc8e70d9417

commit 602316c3cd8556cc78d44f3ea4cd5fc8e70d9417
Author: pbos <pbos@webrtc.org>
Date: Fri Apr 29 23:10:27 2016

Revert of Remove VCMQmRobustness. (patchset #1 id:1 of https://codereview.webrtc.org/1917083003/ )

Reason for revert:
Speculative revert for perf regression.

Original issue's description:
> Remove VCMQmRobustness.
>
> Class contained a lot of not-really-wired-up functionality that ended up
> being complicated ways of saying return 1; or return false;. This
> removes this dependency that complicates code readability significantly.
>
> BUG= webrtc:5066 
> R=marpan@google.com, marpan@webrtc.org
> TBR=stefan@webrtc.org
>
> Committed: https://crrev.com/73894369791cb5eedc8788baf918ec07d11d351d
> Cr-Commit-Position: refs/heads/master@{#12516}

TBR=marpan@webrtc.org,stefan@webrtc.org,marpan@google.com
# Not skipping CQ checks because original CL landed more than 1 days ago.
BUG= webrtc:5066 ,  chromium:607838 

Review-Url: https://codereview.webrtc.org/1935753002
Cr-Commit-Position: refs/heads/master@{#12572}

[modify] https://crrev.com/602316c3cd8556cc78d44f3ea4cd5fc8e70d9417/webrtc/modules/include/module_common_types.h
[modify] https://crrev.com/602316c3cd8556cc78d44f3ea4cd5fc8e70d9417/webrtc/modules/rtp_rtcp/source/producer_fec.cc
[modify] https://crrev.com/602316c3cd8556cc78d44f3ea4cd5fc8e70d9417/webrtc/modules/rtp_rtcp/source/producer_fec_unittest.cc
[modify] https://crrev.com/602316c3cd8556cc78d44f3ea4cd5fc8e70d9417/webrtc/modules/rtp_rtcp/source/rtp_sender_unittest.cc
[modify] https://crrev.com/602316c3cd8556cc78d44f3ea4cd5fc8e70d9417/webrtc/modules/video_coding/media_opt_util.cc
[modify] https://crrev.com/602316c3cd8556cc78d44f3ea4cd5fc8e70d9417/webrtc/modules/video_coding/media_opt_util.h
[modify] https://crrev.com/602316c3cd8556cc78d44f3ea4cd5fc8e70d9417/webrtc/modules/video_coding/media_optimization.cc
[modify] https://crrev.com/602316c3cd8556cc78d44f3ea4cd5fc8e70d9417/webrtc/modules/video_coding/qm_select.cc
[modify] https://crrev.com/602316c3cd8556cc78d44f3ea4cd5fc8e70d9417/webrtc/modules/video_coding/qm_select.h

Even with the blamed CL reverted the regression is active, re-landing.

kjellander@, how can I tell if a bot upgrade or something like that happened in the range? I'm stumped since this is not in Chromium there's no Chromium range affecting it either.

The following revision refers to this bug:
  https://chromium.googlesource.com/external/webrtc.git/+/c04305200ecca57c704b623941a808c03489a20a

commit c04305200ecca57c704b623941a808c03489a20a
Author: pbos <pbos@webrtc.org>
Date: Mon May 02 00:19:05 2016

Reland of move VCMQmRobustness. (patchset #1 id:1 of https://codereview.webrtc.org/1935753002/ )

Reason for revert:
Not root cause for perf regression (regression still ongoing).

Original issue's description:
> Revert of Remove VCMQmRobustness. (patchset #1 id:1 of https://codereview.webrtc.org/1917083003/ )
>
> Reason for revert:
> Speculative revert for perf regression.
>
> Original issue's description:
> > Remove VCMQmRobustness.
> >
> > Class contained a lot of not-really-wired-up functionality that ended up
> > being complicated ways of saying return 1; or return false;. This
> > removes this dependency that complicates code readability significantly.
> >
> > BUG= webrtc:5066 
> > R=marpan@google.com, marpan@webrtc.org
> > TBR=stefan@webrtc.org
> >
> > Committed: https://crrev.com/73894369791cb5eedc8788baf918ec07d11d351d
> > Cr-Commit-Position: refs/heads/master@{#12516}
>
> TBR=marpan@webrtc.org,stefan@webrtc.org,marpan@google.com
> # Not skipping CQ checks because original CL landed more than 1 days ago.
> BUG= webrtc:5066 ,  chromium:607838 
>
> Committed: https://crrev.com/602316c3cd8556cc78d44f3ea4cd5fc8e70d9417
> Cr-Commit-Position: refs/heads/master@{#12572}

TBR=marpan@webrtc.org,stefan@webrtc.org,marpan@google.com
# Not skipping CQ checks because original CL landed more than 1 days ago.
BUG= webrtc:5066 ,  chromium:607838 

Review-Url: https://codereview.webrtc.org/1941643002
Cr-Commit-Position: refs/heads/master@{#12583}

[modify] https://crrev.com/c04305200ecca57c704b623941a808c03489a20a/webrtc/modules/include/module_common_types.h
[modify] https://crrev.com/c04305200ecca57c704b623941a808c03489a20a/webrtc/modules/rtp_rtcp/source/producer_fec.cc
[modify] https://crrev.com/c04305200ecca57c704b623941a808c03489a20a/webrtc/modules/rtp_rtcp/source/producer_fec_unittest.cc
[modify] https://crrev.com/c04305200ecca57c704b623941a808c03489a20a/webrtc/modules/rtp_rtcp/source/rtp_sender_unittest.cc
[modify] https://crrev.com/c04305200ecca57c704b623941a808c03489a20a/webrtc/modules/video_coding/media_opt_util.cc
[modify] https://crrev.com/c04305200ecca57c704b623941a808c03489a20a/webrtc/modules/video_coding/media_opt_util.h
[modify] https://crrev.com/c04305200ecca57c704b623941a808c03489a20a/webrtc/modules/video_coding/media_optimization.cc
[modify] https://crrev.com/c04305200ecca57c704b623941a808c03489a20a/webrtc/modules/video_coding/qm_select.cc
[modify] https://crrev.com/c04305200ecca57c704b623941a808c03489a20a/webrtc/modules/video_coding/qm_select.h

The stats have dropped to values that are a little above the old ones. Can we close the issue or is it needed to look further into this? 

Since reverting the blamed CL did nothing (and there's no new good candidate), and it's comparable to before, I think it's best closed.