1510400 - Promote new tp6-3, 4, 5, and 6 tests to tier 1

Status: RESOLVED FIXED

Description

[Dave Hunt [:davehunt] [he/him] ⌚️UTC]

Once we're satisfied with the results for the tests from bug 1503990 we should promote them to tier 1.

Comment 1

[Robert Wood [:rwood]]

A few things to check/do:

- review how the new test suites (tp6-3, tp6-4, tp6-5, and tp6-6 have been performing as tier 2 on mozilla-central). Are they green consistently or are there issues to be solved first (i.e. frequent intermittent failures).

- does the data look consistent & stable (how noisy is the data compared to the other tp6* suites)

- trigger each new test suite with --gecko-profile and ensure the profiling job works and the resulting profiles can be opened in perf-html.io from the link in treeherder

- update the Raptor wiki with the test details (Bug 1505788)

- add the raw html pages (from playing back each new mitmproxy recording) to the perf-automation repo (also Bug 1505788)

Comment 2

[Robert Wood [:rwood]]

Triggered gecko-profiling jobs on Raptor tp6-3,4,5, and 6 on a central rev here:

https://treeherder.mozilla.org/#/jobs?repo=mozilla-central&searchStr=raptor%2Ctp6&revision=d99bf39f5223abe554a629a947eee344c3e9e29e

Comment 3

[Robert Wood [:rwood]]

I'd like to land a fix for intermittent Bug 1502032 first which is just a [taskcluster:error] Task aborted - max run time exceeded (need to increase the overall job time).

Also note that these new jobs (and other existing tp6-1 and tp6-2 jobs) on google chrome linux64 fail; this is a known issue (Bug 1495903) but we always leave Raptor google chrome jobs as tier 2 anyway, so that won't prevent us from moving tier 1 for Firefox.

Comment 4

[Robert Wood [:rwood]]

Looking at data noise for these new Raptor tp6-3, 4, 5, and 6 jobs on mozilla-central: Bing (tp6-4) and Reddit (tp6-6) seem to fluctuate the most, at least from the limited numbers I looked at.

:jmaher, can you please remind me how I can get the noise factor for specific test jobs? I want to be sure Bing and Reddit aren't too noisy for automated regression detection.

If I do a perfherder compare of one mozilla-central job with others from the last 2 days (perfherder default) and narrow it down to 'raptor-tp6-bing-firefox' there's no noise factor available. Maybe this isn't the right method...

https://treeherder.mozilla.org/perf.html#/compare?originalProject=mozilla-central&newProject=mozilla-central&newRevision=d99bf39f5223abe554a629a947eee344c3e9e29e&framework=10&filter=raptor-tp6-bing-firefox&selectedTimeRange=172800

Comment 5

[Joel Maher ( :jmaher ) (UTC-4)]

I compare against the same build- typically 6+ data points for the revision,
I have retriggered for this build:
https://treeherder.mozilla.org/#/jobs?repo=mozilla-central&searchStr=tp6%2Cfirefox&group_state=expanded&revision=a89d378954538f7fe0cad49681b409e80c3f8a0f&selectedJob=214657334

and in the near future we will see data to look at here:
https://treeherder.mozilla.org/perf.html#/compare?originalProject=mozilla-central&originalRevision=a89d378954538f7fe0cad49681b409e80c3f8a0f&newProject=mozilla-central&newRevision=a89d378954538f7fe0cad49681b409e80c3f8a0f&framework=1

Comment 6

[Robert Wood [:rwood]]

Thanks :jmaher for the compare view link!

https://treeherder.mozilla.org/perf.html#/compare?originalProject=mozilla-central&originalRevision=a89d378954538f7fe0cad49681b409e80c3f8a0f&newProject=mozilla-central&newRevision=a89d378954538f7fe0cad49681b409e80c3f8a0f&framework=10&filter=tp6&showOnlyComparable=1

I don't see anything too concerning, a few high standard deviations on some configurations but they mostly look like outliers to me. I think we should go ahead and upgrade these to tier 1.

Comment 7

[Robert Wood [:rwood]]

Attachment: Bug 1510400 - Promote new tp6-3, 4, 5, and 6 tests to tier 1; r?jmaher,davehunt

Comment 8

[Robert Wood [:rwood]]

https://treeherder.mozilla.org/#/jobs?repo=try&revision=f328404243d3a0cdf163d7e079475df0185cfa96

Comment 9

[Pulsebot]

Pushed by [e-mail]:
https://hg.mozilla.org/integration/autoland/rev/a2d7fd0f5139
Promote new tp6-3, 4, 5, and 6 tests to tier 1; r=jmaher

Comment 10

[Eliza Balazs [:ebalazs_]]

https://hg.mozilla.org/mozilla-central/rev/a2d7fd0f5139