Hey Ethan, Is there any update on this?

Wrote a comment on the GitHub issue; I have a fix that turned out to be not quite correct but has a complement on the bisect side that will make it work. I'll implement it early next week which should fix this issue.

The following revision refers to this bug:
  https://chromium.googlesource.com/chromium/tools/build.git/+/cee7475ca3ab8eba9c9760cdf086cc7c703df15c

commit cee7475ca3ab8eba9c9760cdf086cc7c703df15c
Author: eakuefner <eakuefner@chromium.org>
Date: Tue Sep 20 22:25:10 2016

[Bisect] Account for tir_label-level summaries in parse_metric

This should fix bisects for PCv2, which are broken for tir_label summaries,
which we didn't used to output.

BUG= 644865 

Review-Url: https://codereview.chromium.org/2352203002

[modify] https://crrev.com/cee7475ca3ab8eba9c9760cdf086cc7c703df15c/scripts/slave/recipe_modules/auto_bisect/config_validation.py
[modify] https://crrev.com/cee7475ca3ab8eba9c9760cdf086cc7c703df15c/scripts/slave/recipe_modules/auto_bisect/config_validation_test.py
[modify] https://crrev.com/cee7475ca3ab8eba9c9760cdf086cc7c703df15c/scripts/slave/recipe_modules/bisect_tester/parse_metric.py

The following revision refers to this bug:
  https://chrome-internal.googlesource.com/chrome/tools/build_limited/scripts/slave/+/c4403d55c2cdb3d3a0aea47ea9d5f0f48071f76d

commit c4403d55c2cdb3d3a0aea47ea9d5f0f48071f76d
Author: recipe-roller <recipe-roller@chromium.org>
Date: Tue Sep 20 22:34:14 2016

The following revision refers to this bug:
  https://chromium.googlesource.com/chromium/src.git/+/1fbd04e06a34b13dc9659927efb4a19d6a978a9b

commit 1fbd04e06a34b13dc9659927efb4a19d6a978a9b
Author: recipe-roller <recipe-roller@chromium.org>
Date: Tue Sep 20 22:58:18 2016

Roll recipe dependencies (trivial).

This is an automated CL created by the recipe roller. This CL rolls recipe
changes from upstream projects (e.g. depot_tools) into downstream projects
(e.g. tools/build).

More info is at https://goo.gl/zkKdpD. Use https://goo.gl/noib3a to file a bug
(or complain)

build:
  https://crrev.com/cee7475ca3ab8eba9c9760cdf086cc7c703df15c [Bisect] Account for tir_label-level summaries in parse_metric (eakuefner@chromium.org)

TBR=martiniss@chromium.org,phajdan.jr@chromium.org
BUG= 644865 

Recipe-Tryjob-Bypass-Reason: Autoroller
Bugdroid-Send-Email: False
Review-Url: https://codereview.chromium.org/2354113002
Cr-Commit-Position: refs/heads/master@{#419883}

[modify] https://crrev.com/1fbd04e06a34b13dc9659927efb4a19d6a978a9b/infra/config/recipes.cfg

The following revision refers to this bug:
  https://chromium.googlesource.com/infra/infra.git/+/db8d44b26c5caaa7105b87e3c36abd876841f4de

commit db8d44b26c5caaa7105b87e3c36abd876841f4de
Author: recipe-roller <recipe-roller@chromium.org>
Date: Tue Sep 20 23:05:53 2016

Roll recipe dependencies (trivial).

This is an automated CL created by the recipe roller. This CL rolls recipe
changes from upstream projects (e.g. depot_tools) into downstream projects
(e.g. tools/build).

More info is at https://goo.gl/zkKdpD. Use https://goo.gl/noib3a to file a bug
(or complain)

build:
  https://crrev.com/9672983b7e4143ea26754935e0132a0838f2b326 Add flutter_services (abarth@chromium.org)
  https://crrev.com/cee7475ca3ab8eba9c9760cdf086cc7c703df15c [Bisect] Account for tir_label-level summaries in parse_metric (eakuefner@chromium.org)

TBR=martiniss@chromium.org,phajdan.jr@chromium.org
BUG= 644865 

Recipe-Tryjob-Bypass-Reason: Autoroller
Bugdroid-Send-Email: False
Review-Url: https://codereview.chromium.org/2356043002

[modify] https://crrev.com/db8d44b26c5caaa7105b87e3c36abd876841f4de/infra/config/recipes.cfg

I pushed Ethan's changes live to the dashboard and kicked off https://chromeperf.appspot.com/buildbucket_job_status/9000900884809620160.

Also kicked https://chromeperf.appspot.com/buildbucket_job_status/9000922468572912368 on Windows which may complete a little faster. I lowered the repeat count as well so we can get some quick feedback about whether this fix works.

Whoops, I accidentally kicked it from the default version, so mine failed. We're still waiting on the one Annie kicked above to come back, which should hopefully be soon. Once that happens I can go through the deployment checklist and set as default, as long as the fix works.

It's promising that timeToFirstContentfulPaint_avg/pcv1-cold/http___jugem.jp_ is the metric being analyzed, which should work fine for bisect.

Something is off about Annie's bisect -- it seems that the performance test is hung. However, it passed the "Get Reference Values" step, which is where my faulty bisect stopped, so it seems like my fix works.

Annie, can you set the version you deployed this morning as default?

We've been having oom issues on all recent deployments of the Dashboard, which we believe just got fixed. Current default version is tot as of about an hour ago. 

Okay, the version with the fix should be the default now. I saw Kari kicked off some bisects yesterday, I will kick off more.

A few bisects:
https://chromeperf.appspot.com/buildbucket_job_status/9000723249222065152
https://chromeperf.appspot.com/buildbucket_job_status/9000723210014951456

Issue 643638 has been merged into this issue.

Issue 643637 has been merged into this issue.

As bug triager, I've made sure that each bug that blocks on this one has had bisects re-kicked since the outage was resolved.

This bug can be closed.