Right now the pageload tests for Fenix get hidden on try because they are tier-3. We should make some changes to set all selected tasks in mach try perf to tier-2 and higher so that users won't be wondering where they are.

@sparky do we know why the Fenix tests are running as tier 3? Maybe we can upgrade them? It may still make sense for mach try perf to bump tiers up for convenience.

Good question, I think it might be an artifact from the browsertime migration. We set the mobile tests to tier-3 by default here: https://searchfox.org/mozilla-central/source/taskcluster/ci/test/browsertime-mobile.yml#11
Which comes from this bug: https://bugzilla.mozilla.org/show_bug.cgi?id=1585120

I wonder if we could remove refbrow at the same time as updating the tier default setting. :davehunt what do you think?

[an Android engineer here] I would be cautious about making them tier 1 (unless all the tests are currently firmly green).

Also, a note on the reference browser: I think it might be useful to keep it, as it is a smaller app that uses Android Components, so should we have a degradation in performance, we can trace the problematic layer better (is it Fenix itself? Is it Android Components? Is it Gecko/GeckoView?). So I see having all three apps beneficial.

Clearing the needinfo based on the feedback from @owlish.

(In reply to [:owlish] 🦉 PST from comment #5)

Also, a note on the reference browser: I think it might be useful to keep it, as it is a smaller app that uses Android Components, so should we have a degradation in performance, we can trace the problematic layer better (is it Fenix itself? Is it Android Components? Is it Gecko/GeckoView?). So I see having all three apps beneficial.

Sounds good, something to note is that we don't run any tests on refbrow at the moment. It's only available for try runs.