The right person is me; I'm looking into this right now.

Okay, so it turns out that this is yet another subtle bug in Telemetry's summarization system.

For context, remember that in TBMv1 we did not merge across IRs; we viewed different interaction records as non-comparable. But now, we've overloaded the tir_label concept to more simply represent a generic label, as in this case.

I'm just going to patch it so that we do merge across IRs if it's a TBMv2 benchmark, since Ben will be implementing summarization in TBMv2.1 so we will probably anyway be able to switch away from Telemetry's built-in summarization and remove the hack soonish.

Looking at this today, I think the fix is actually easier than anticipated: we simply need to do two summarization passes in TBMv2: one that keys both by name and by tir_label (the default behavior), and another that keys only by name (so that even values with different tir_labels will be merged).

This implies that there are actually two different types of values missing here, which might be masked if the memory system health benchmarks do not use page repeat or page set repeat: per-page values from across all tir_labels, and summary values from across all tir_labels. Summary produces both per-page and summary values, which is why I think a two-pass summary is the right approach.

Working on the CL for this, and it turns out that without the second summarization pass, there are no non-per-page, non-tir-label-specific values emitted in the chart JSON, as hypothesized! This is based on a unit test I wrote to test the new functionality, that I also tried with should_flatten_tir_labels set to False to see what would happen.

I'm confident based on this that I have the correct approach. CL incoming shortly.

Okay, as a final update for the day, this approach does work, but it turns out to be slightly more of a headache just from a refactoring standpoint because I need to update all the merging routines for all the value types, and their associated tests. I'll end up doing that tomorrow morning.

The following revision refers to this bug:
  https://chromium.googlesource.com/chromium/src.git/+/eaed87169fe72b9c0a8a8c749856dd8a9d7c12ba

commit eaed87169fe72b9c0a8a8c749856dd8a9d7c12ba
Author: catapult-deps-roller <catapult-deps-roller@chromium.org>
Date: Wed Jul 13 05:06:01 2016

Roll src/third_party/catapult/ 72fb0b506..65c2d9df0 (4 commits).

https://chromium.googlesource.com/external/github.com/catapult-project/catapult.git/+log/72fb0b5062af..65c2d9df0d06

$ git log 72fb0b506..65c2d9df0 --date=short --no-merges --format='%ad %ae %s'

BUG= 626979 , 624522 

TBR=catapult-sheriff@chromium.org

Review-Url: https://codereview.chromium.org/2147543004
Cr-Commit-Position: refs/heads/master@{#405037}

[modify] https://crrev.com/eaed87169fe72b9c0a8a8c749856dd8a9d7c12ba/DEPS

The following revision refers to this bug:
  https://chromium.googlesource.com/chromium/src.git/+/eaed87169fe72b9c0a8a8c749856dd8a9d7c12ba

commit eaed87169fe72b9c0a8a8c749856dd8a9d7c12ba
Author: catapult-deps-roller <catapult-deps-roller@chromium.org>
Date: Wed Jul 13 05:06:01 2016

Roll src/third_party/catapult/ 72fb0b506..65c2d9df0 (4 commits).

https://chromium.googlesource.com/external/github.com/catapult-project/catapult.git/+log/72fb0b5062af..65c2d9df0d06

$ git log 72fb0b506..65c2d9df0 --date=short --no-merges --format='%ad %ae %s'

BUG= 626979 , 624522 

TBR=catapult-sheriff@chromium.org

Review-Url: https://codereview.chromium.org/2147543004
Cr-Commit-Position: refs/heads/master@{#405037}

[modify] https://crrev.com/eaed87169fe72b9c0a8a8c749856dd8a9d7c12ba/DEPS

Ethan: ping.

This overall values are still not reported: https://chromeperf.appspot.com/report?sid=c2b03d24bb7aa777f9db6695b49f50b8141d78373e5c9f970bdad99cc892a5e2

Petr, sorry for the latency. As an update, I just need to make some minor fixes to https://codereview.chromium.org/2142293002 and then it should be ready for Ned to stamp.

Petr, as a final update, I've designated Fridays to work on this and other outstanding Telemetry issues, so I expect to get to this at that time.

Petr, https://codereview.chromium.org/2142293002/ has landed in Catapult so these numbers should begin showing up on the perf dashboard. I'm going to mark this bug fixed; kindly verify.

The following revision refers to this bug:
  https://chromium.googlesource.com/chromium/src.git/+/d9c3aae54d0cb2fd920a6ad4727d70a38c72057d

commit d9c3aae54d0cb2fd920a6ad4727d70a38c72057d
Author: catapult-deps-roller <catapult-deps-roller@chromium.org>
Date: Mon Aug 15 21:51:03 2016

Roll src/third_party/catapult/ 1c6ff37b8..1e3f9b906 (4 commits).

https://chromium.googlesource.com/external/github.com/catapult-project/catapult.git/+log/1c6ff37b8350..1e3f9b906996

$ git log 1c6ff37b8..1e3f9b906 --date=short --no-merges --format='%ad %ae %s'

BUG= 624522 

TBR=catapult-sheriff@chromium.org

Review-Url: https://codereview.chromium.org/2252463002
Cr-Commit-Position: refs/heads/master@{#412058}

[modify] https://crrev.com/d9c3aae54d0cb2fd920a6ad4727d70a38c72057d/DEPS