The following revision refers to this bug:
  https://chromium.googlesource.com/chromium/tools/build/+/1d125357f77f2747805c5739589bcff7374ccd38

commit 1d125357f77f2747805c5739589bcff7374ccd38
Author: Patrik Höglund <phoglund@chromium.org>
Date: Wed Nov 15 09:05:15 2017

Split browser tests into functional and perf tests.

This enables the functional tests to run with the default timeout
rather than the insane timeout required by the video quality tests.
This way, if something causes all tests to hang it won't take hours
for the bot to time out.

Next patch: try if the functional tests can run in parallel after
all.

Bug:chromium:784831
Change-Id: Ib5e7dacd71eaf6f6f814886b99509e8849ad5da0
Reviewed-on: https://chromium-review.googlesource.com/768680
Reviewed-by: Sergiy Byelozyorov <sergiyb@chromium.org>
Commit-Queue: Patrik Höglund <phoglund@chromium.org>

[modify] https://crrev.com/1d125357f77f2747805c5739589bcff7374ccd38/scripts/slave/recipe_modules/chromium_tests/chromium_webrtc.py
[modify] https://crrev.com/1d125357f77f2747805c5739589bcff7374ccd38/scripts/slave/recipe_modules/chromium_tests/steps.py

Next step: try running the functional tests in parallel

Should we move the perf tests to their own builders? So they don't block LKGR?

Yeah. The perf tests take 19 minutes to run for 17 tests. The perf tests aren't much use for LKGR purposes. The functional tests take about 2 minutes, so parallellizing those won't bring much gain.

It is nice to run perf tests on LKGR bots because we get a signal from WebRTC before it rolls into Chromium. That advantage has diminished now that rolls happen more often, though.

We could start imagining a new bot architecture, but I think we should do that in Q1 in that case. We could probably make the chromium waterfalls into VMs and create a perf+baremetal waterfall which also runs the webcam tests. In that case we might as well drop the .chromium wf since it does the same thing as regular Chrome bots (we'd just keep the FYI wf). This shouldn't make a difference for the AppRTC or webcam tests; they're not on trybots anyway, so moving to a different wf doesn't make a difference for them as long as the new perf+baremetal wf is clearly visible like the old one.

I actually only see advantages with this architecture, but let's think more about that...

The following revision refers to this bug:
  https://chromium.googlesource.com/chromium/tools/build/+/44de4837ab5ce019d1df5247d42ec93a7a592907

commit 44de4837ab5ce019d1df5247d42ec93a7a592907
Author: Patrik Höglund <phoglund@chromium.org>
Date: Wed Nov 15 16:01:58 2017

browser_tests_perf -> browser_tests.

I realized it's a bad idea to rename the target for the perf tests,
as we'd have to migrate data in the perf dashboard. Let's create
a new name for the functional tests instead.

TBR=sergiyb@chromium.org
Bug:chromium:784831
Change-Id: I91b3324082feda0e4a6d199fc849032a34303213
Reviewed-on: https://chromium-review.googlesource.com/771632
Reviewed-by: Patrik Höglund <phoglund@chromium.org>
Reviewed-by: Edward Lesmes <ehmaldonado@chromium.org>
Commit-Queue: Patrik Höglund <phoglund@chromium.org>

[modify] https://crrev.com/44de4837ab5ce019d1df5247d42ec93a7a592907/scripts/slave/recipe_modules/chromium_tests/chromium_webrtc.py
[modify] https://crrev.com/44de4837ab5ce019d1df5247d42ec93a7a592907/scripts/slave/recipe_modules/chromium_tests/steps.py

With all that said, maybe we can make things a bit better in the short term by cutting down the amount of tests. I think the problem is that we're running 60-sec tests for a lot of permutations, like several codecs, several resolutions etc. Maybe we can select the most important permutations and throw others away.

The following revision refers to this bug:
  https://chromium.googlesource.com/chromium/tools/build/+/dd3fb217f15331370c644c907731b564051defc0

commit dd3fb217f15331370c644c907731b564051defc0
Author: Edward Lesmes <ehmaldonado@chromium.org>
Date: Wed Nov 15 16:26:10 2017

Revert "browser_tests_perf -> browser_tests."

This reverts commit 44de4837ab5ce019d1df5247d42ec93a7a592907.

Reason for revert: Breaks bots

Original change's description:
> browser_tests_perf -> browser_tests.
> 
> I realized it's a bad idea to rename the target for the perf tests,
> as we'd have to migrate data in the perf dashboard. Let's create
> a new name for the functional tests instead.
> 
> TBR=sergiyb@chromium.org
> Bug:chromium:784831
> Change-Id: I91b3324082feda0e4a6d199fc849032a34303213
> Reviewed-on: https://chromium-review.googlesource.com/771632
> Reviewed-by: Patrik Höglund <phoglund@chromium.org>
> Reviewed-by: Edward Lesmes <ehmaldonado@chromium.org>
> Commit-Queue: Patrik Höglund <phoglund@chromium.org>

TBR=phoglund@chromium.org,sergiyb@chromium.org,ehmaldonado@chromium.org

Change-Id: I2040ea9d433a509276128c8ae10ac7bb8f5e003b
No-Presubmit: true
No-Tree-Checks: true
No-Try: true
Bug: chromium:784831
Reviewed-on: https://chromium-review.googlesource.com/771714
Reviewed-by: Edward Lesmes <ehmaldonado@chromium.org>
Commit-Queue: Edward Lesmes <ehmaldonado@chromium.org>

[modify] https://crrev.com/dd3fb217f15331370c644c907731b564051defc0/scripts/slave/recipe_modules/chromium_tests/chromium_webrtc.py
[modify] https://crrev.com/dd3fb217f15331370c644c907731b564051defc0/scripts/slave/recipe_modules/chromium_tests/steps.py

The following revision refers to this bug:
  https://chromium.googlesource.com/chromium/tools/build/+/d43f60d049cf7fd6352cd20cbe1859f01fee4c7e

commit d43f60d049cf7fd6352cd20cbe1859f01fee4c7e
Author: Edward Lemur <ehmaldonado@chromium.org>
Date: Wed Nov 15 16:40:28 2017

Reland "browser_tests_perf -> browser_tests."

This is a reland of 44de4837ab5ce019d1df5247d42ec93a7a592907
Original change's description:
> browser_tests_perf -> browser_tests.
>
> I realized it's a bad idea to rename the target for the perf tests,
> as we'd have to migrate data in the perf dashboard. Let's create
> a new name for the functional tests instead.
>
> TBR=sergiyb@chromium.org
> Bug:chromium:784831
> Change-Id: I91b3324082feda0e4a6d199fc849032a34303213
> Reviewed-on: https://chromium-review.googlesource.com/771632
> Reviewed-by: Patrik Höglund <phoglund@chromium.org>
> Reviewed-by: Edward Lesmes <ehmaldonado@chromium.org>
> Commit-Queue: Patrik Höglund <phoglund@chromium.org>

TBR=sergiyb@chromium.org
Bug: chromium:784831
Change-Id: I746bb94198c65bd3d825252980da565c36d3b1de
Reviewed-on: https://chromium-review.googlesource.com/771715
Reviewed-by: Edward Lesmes <ehmaldonado@chromium.org>
Commit-Queue: Edward Lesmes <ehmaldonado@chromium.org>

[modify] https://crrev.com/d43f60d049cf7fd6352cd20cbe1859f01fee4c7e/scripts/slave/recipe_modules/chromium_tests/chromium_webrtc.py
[modify] https://crrev.com/d43f60d049cf7fd6352cd20cbe1859f01fee4c7e/scripts/slave/recipe_modules/chromium_tests/steps.py

target_name, of course. Thanks for fixing Edward!

No problem :)
Sorry I didn't catch it on the review

Assigning this to you, since you last worked on this.

-> Oleh, since restructuring waterfalls is on his OKRs.

 Issue 815098  has been merged into this issue.