1122066 - Add proper daily collection mechanism

Status: RESOLVED INVALID

(Firefox Health Report Graveyard :: Client: Desktop, defect)

Description

[Georg Fritzsche [:gfritzsche]]

Currently telemetry histograms etc. are mostly push-based (i.e. client pushes data to telemetry when it has new values).

We may need to add a proper pull mechanism for data collection from clients on telemetry main ping generation (i.e. at 24h intervals, environment changes, shutdown).

Comment 1

[Georg Fritzsche [:gfritzsche]]

* observers are not great, we may want to do async / promise-based
* clients may register js listeners?

Comment 2

[Benjamin Smedberg]

I think there's a disconnect here in that we're often using histograms to record simple values. Histograms really shouldn't need to be pulled at any particular point in time, while we should collect a set of simple values together at appropriate points.

Comment 3

[Georg Fritzsche [:gfritzsche]]

We do currently have use-cases like "daily" measurements that we want to port over.
We may also want to do things like SEARCH_DEFAULT_ENGINE per "sub session" (i.e. on day or environment changes).

Whether we trigger histograms or other things, we need a proper trigger mechanism.

Comment 4

[Georg Fritzsche [:gfritzsche]]

We'll investigate the situation first in bug 1127907, then file follow-up bugs with actual actions from there.