Working on reproduce this...

I am able to reproduce and grab a passing trace (with stable Chrome), failing trace (with Chrome build in nasko@'s CL) in it.

Deleted: pass_trace.html 4.0 MB

pass_trace.html 4.0 MB View Download

Deleted: fail_trace.html 4.0 MB

fail_trace.html 4.0 MB View Download

We can see that telemetry.internal.warm_cache.warm.start is in the pass_trace.html (Renderer (pid 11036)) but not fail_trace.html

Also the fact that there are two renderer processes in fail_trace.html confirm my theory that this test is broken because the trace event start call & trace event end call now live in two different processes, thus breaking the trace event.

Hmhh, actually looking at this carefully, the problem is different from what I originally thought.

We *do* create 2 trace markers before & after the cache warm up:
"start" marker in https://cs.chromium.org/chromium/src/third_party/catapult/telemetry/telemetry/page/cache_temperature.py?rcl=5b2cfb7fa5f197e561c81012a7cc17e5c19c84ec&l=45

"end" marker in https://cs.chromium.org/chromium/src/third_party/catapult/telemetry/telemetry/page/cache_temperature.py?rcl=5b2cfb7fa5f197e561c81012a7cc17e5c19c84ec&l=54

Looking at the trace, these markers only take 1.5ms-ish due to them being two continuous js calls from Telemetry.

So this is the problem of some how the "start" trace event cannot be created in the renderer process of "about:blank" page with nasko's change. 

Ok, I think I know what's going on. The order of the events are as follows:
  Start at about://blank
  Issue telemetry.internal.warm_cache.warm.start
  Navigate to http://google.com (which does not exists)
  Issue telemetry.internal.warm_cache.warm.end
  Issue telemetry.internal.warm_cache.hot.start
  Navigate to http://google.com (which does not exists)
  Issue telemetry.internal.warm_cache.hot.end

So because the page google.com doesn't have associated archive with it in the test, it is a an error page. nasko's CL make a new process with that error page, thus killing the old renderer process of "about:blank" page, making us losing all the trace events of that page.

This is a bug due to the fact that tracing v1 isn't able to keep the trace events from dead processes during recording. I will try to make the test use a valid page and hope that unblock nasko@'s CL.

Ok, use a working page fixed the test! I will make the patch fix tomorrow.

The following revision refers to this bug:
  https://chromium.googlesource.com/catapult/+/754d7301993f8915adb53c3b2d85b9c587fb2865

commit 754d7301993f8915adb53c3b2d85b9c587fb2865
Author: nednguyen <nednguyen@google.com>
Date: Thu May 03 17:45:44 2018

Fix cache_temperature_unittes to use working pages from test archives

Bug:  chromium:839127 
Change-Id: I72d2a690d12ab24473f595b4a3c91acc714f0312
TBR=perezju@chromium.org
Reviewed-on: https://chromium-review.googlesource.com/1040989
Reviewed-by: Ned Nguyen <nednguyen@google.com>
Commit-Queue: Ned Nguyen <nednguyen@google.com>

[modify] https://crrev.com/754d7301993f8915adb53c3b2d85b9c587fb2865/telemetry/telemetry/internal/testing/test_page_sets/data/example_domain.json
[modify] https://crrev.com/754d7301993f8915adb53c3b2d85b9c587fb2865/telemetry/telemetry/internal/testing/test_page_sets/example_domain.py
[modify] https://crrev.com/754d7301993f8915adb53c3b2d85b9c587fb2865/telemetry/telemetry/page/cache_temperature_unittest.py

The following revision refers to this bug:
  https://chromium.googlesource.com/chromium/src.git/+/f3cb54067e379373eebb8a58399e36f9046a7c35

commit f3cb54067e379373eebb8a58399e36f9046a7c35
Author: catapult-chromium-autoroll@skia-buildbots.google.com.iam.gserviceaccount.com <catapult-chromium-autoroll@skia-buildbots.google.com.iam.gserviceaccount.com>
Date: Fri May 04 00:44:42 2018

Roll src/third_party/catapult/ 19282cf9d..153134ef2 (4 commits)

https://chromium.googlesource.com/catapult.git/+log/19282cf9d32d..153134ef26c3

$ git log 19282cf9d..153134ef2 --date=short --no-merges --format='%ad %ae %s'
2018-05-03 dtu [pinpoint] Increase task queue rate.
2018-05-03 szager Add support for profiling on Android using simpleperf
2018-05-02 ssid Fix symbolization of heap profiles when so is mapped from apk
2018-05-03 nednguyen Fix cache_temperature_unittes to use working pages from test archives

Created with:
  roll-dep src/third_party/catapult
BUG=chromium:734705, chromium:839127 


The AutoRoll server is located here: https://catapult-roll.skia.org

Documentation for the AutoRoller is here:
https://skia.googlesource.com/buildbot/+/master/autoroll/README.md

If the roll is causing failures, please contact the current sheriff, who should
be CC'd on the roll, and stop the roller if necessary.


TBR=sullivan@chromium.org

Change-Id: I97dd890c7eda177ae8b603b830e60bfbb215da98
Reviewed-on: https://chromium-review.googlesource.com/1043079
Commit-Queue: catapult-chromium-autoroll <catapult-chromium-autoroll@skia-buildbots.google.com.iam.gserviceaccount.com>
Reviewed-by: catapult-chromium-autoroll <catapult-chromium-autoroll@skia-buildbots.google.com.iam.gserviceaccount.com>
Cr-Commit-Position: refs/heads/master@{#555933}
[modify] https://crrev.com/f3cb54067e379373eebb8a58399e36f9046a7c35/DEPS