I think the best place to follow up the discussion in https://github.com/catapult-project/catapult/issues/2722

The fact that we don't monitor all the metrics is unfortunate but WAI. We did that until a while ago and that caused the sheriffs to riot against memory-infra :)
The problem is that when you have an alert in a sub-metric (say webcache) that also fires recursively alerts for (browser, renderer, all processes) x (webcache, private_dirty, partition_alloc, native_heap, pss)
so you get easily 12x the amount of alerts you should need.
There is a long term plan to express an "influencing graph", at the metric level, to group metrics but that's not there yet.

Going back to the quesiton here "why [1] did not trigger an alert", perezju explained me that the problem is that according to the current alerts config (see catapult bug) we monitor only averages per story-group and not per story (/* vs /*/* in the alert configs). So, very likely that flipboard alert got diluted by the other stories not regressing.

The latter (story vs story group alerts) is probably a genuine mistake and not WAI, and we should put it back on individual stories.

(Also, we need a better UI story for the dashboard, as navigating in my own metrics is becoming impossible. but this is another story.)

[1] https://chromeperf.appspot.com/report?sid=24541df628a91a62504b0954e454a7483e6058a8c1c78713c3dfd473a7fd017a&start_rev=405220&end_rev=424328

Annie is out, so reassign this to the Automation sheriff

Sorry for the long delay here, should this be closed in favor of https://github.com/catapult-project/catapult/issues/2722 ?

I would say yes. Note that skia in particular is now included in the list of alerted metrics.

Please send any further comments to the catapult bug, reopening if needed.