Issue metadata
Sign in to add a comment
|
A zero-to-nonzero to 14.6% regression in webrtc_perf_tests at 15037:15038 |
||||||||||||||||||||
Issue descriptionI suspect this CL: https://chromium.googlesource.com/external/webrtc/+/e6f98c7a379aae970e7570ac3cf99e2a21f256c0 The other CL in the blame range is an automatic chrome roll.
,
Nov 16 2016
Yes, that seems very plausible. My first guess is that these tests rely on undocumented behaviour: previously RED over RTX was implicitly enabled whenever RED was enabled. After the suspected CL, this is no longer the case. I'll look into it in detail.
,
Nov 17 2016
The following revision refers to this bug: https://chromium.googlesource.com/external/webrtc.git/+/fbfb536ee94dcebfc2f80ccfea1bbab2b1a4cd69 commit fbfb536ee94dcebfc2f80ccfea1bbab2b1a4cd69 Author: brandtr <brandtr@webrtc.org> Date: Thu Nov 17 12:18:37 2016 Explicitly enable RED over RTX in rampup tests. Also remove unused |rtx_ssrc_map_| member. BUG= chromium:665923 Review-Url: https://codereview.webrtc.org/2508973002 Cr-Commit-Position: refs/heads/master@{#15127} [modify] https://crrev.com/fbfb536ee94dcebfc2f80ccfea1bbab2b1a4cd69/webrtc/call/rampup_tests.cc [modify] https://crrev.com/fbfb536ee94dcebfc2f80ccfea1bbab2b1a4cd69/webrtc/call/rampup_tests.h
,
Nov 17 2016
,
Nov 17 2016
The metrics are back up in WebRTC revision #15127. There will be a bunch of alerts in a day or two :)
,
Nov 21 2016
Added some of the expected alerts to this bug. |
|||||||||||||||||||||
►
Sign in to add a comment |
|||||||||||||||||||||
Comment 1 by hlundin@chromium.org
, Nov 16 2016