brendan c Reporter
	Description • 9 years ago

from email in fhr dev:

I just discovered a top level "errors" field in an FHR v2 payload that
I was looking at. I've never seen this field before (it's not
mentioned in the docs at:
https://hg.mozilla.org/mozilla-central/file/600f44fd317c/services/healthreport/docs/dataformat.rst
). This field seems to have a stack trace of whatever problem caused
the client to fail to submit any data.days or data.last information.

Will a similar error reporting exist in v4? If so, we should check
these error reports on ping reception (maybe as part of the monitoring
happening in https://bugzilla.mozilla.org/show_bug.cgi?id=1129185 , or
a new bug if needed) and these pings should be set aside and directed
to Georg and the other folks working on the client side of things. If
we won't have this kind of error reporting for v4, maybe we should add
it back in?

(Sorry to have not mentioned this before-- like I said, I didn't know
this existed in v2, and as a non-engineer, self-reporting of errors
didn't occur to me).

Also, it's a little late in the v2 game at this point, but did anyone
ever look at these errors for v2? Prevalence, patterns, etc? Maybe
there is something to learn for v4...

	Georg Fritzsche [:gfritzsche]
	Comment 1 • 9 years ago

We won't have this reporting in the current Telemetry implementation, but Vladan wants us to have improved Telemetry error reporting anyway.

From the sound of it, this will not be useful to port over directly (as it is FHR implementation specific), so we should look into proper improved Telemetry diagnostics later.

	Benjamin Smedberg
	Comment 2 • 9 years ago

I agree this is not a requirement right now. I think we should have a separate project to have better error reporting from the browser in general (not just from the telemetry code) but that will probably be a separate ping type and require significant design/privacy work to get right.

	Vladan Djeric (:vladan)
	Comment 3 • 9 years ago

Should we revisit this topic?

	Georg Fritzsche [:gfritzsche]
	Comment 4 • 9 years ago

I'm not sure what we would currently gain from it - we are checking into the diagnostics again and whether the cover the data quality questions we have.
Until then i don't know where capturing stack traces would be helpful.

	Georg Fritzsche [:gfritzsche]
	Comment 5 • 9 years ago

We have a potential more general project for a Telemetry error reporting coming up.
Closing this bug out over that project.