1386597 - Investigate differences between bsmedberg/daily-latency-metrics, error_aggregates, and quantum_rc

Status: RESOLVED FIXED

(Data Platform and Tools :: General, enhancement, P1)

Description

[Frank Bertsch [:frank]]

These differences need to be explained, and we need consensus on which values are "correct".

Comment 1

[Mauro Doglio [:mdoglio]]

Just adding a data point here, I checked data from yesterday (usage_hours, number of pings processed, main and content crashes) on both crash_aggregates and error_aggregates and the numbers seems to be in the same ballpark.

Comment 2

[Frank Bertsch [:frank]]

I've noted a few differences between bsmedberg's script and error_aggregates:
1. bsmedberg limits to just Windows_NT clients, while error_aggregates has os_name as a dimension
2. bsmedberg limits to clients with e10sEnabled, while error_aggregates has quantum_ready, which indicates the client has e10sEnabled. (Note that quantum_ready = newstyle theme, no addons [i.e. they are system or webextensions only], and e10sEnabled)

These differences do not explain all of the count differences between the two scripts, but only for the beginning of the period [0].

[0] https://screenshots.firefox.com/ubwoHr0Mu7QruIl2/localhost

Comment 3

[Frank Bertsch [:frank]]

Just to follow up on :mdoglio's point, see https://sql.telemetry.mozilla.org/queries/11807/source and https://sql.telemetry.mozilla.org/queries/11806/source.