1356837 - Review GPU_PROCESS_CRASH_FALLBACKS expiry and add test around its collection

Status: RESOLVED WONTFIX

(Core :: Audio/Video: Playback, enhancement, P3)

Description

[Gerald Squelart (he/him) (not at Mozilla since 2022-09-15)]

Spawned from bug 1338011, which added an intermediate software fallback for compositor process crashes and a related telemetry histogram.

The histogram was initially set to never expire, but :bsmedberg pointed out in bug 1338011 comment 25 that:
> It doesn't appear that there is any automated test which verifies that this
> telemetry is being recorded correctly. Typically that is required for
> expires: never telemetry, because it's common/easy to accidentally break
> this kind of telemetry over time. Please at least file a followup bug to add
> automation for this.
> 
> Who is responsible for monitoring this data? Note that automated alerts will
> not be useful in this case, nor t.m.o, since those only operate on
> prerelease/opt-in datasets. Is this something that we should have as part of
> project war room [1] for permanent monitoring as part of release health?

After this, and due to a change of bug owner and short time before a new release, the histogram was just changed to expire in 60, so we don't just unleash an untested never-expiring histogram.

Hence this follow-up bug.
- Who will be responsible for monitoring the data?
- Can we create an automated test to verify the correct recording of telemetry?
- Suitability for "Project War Room [1]"?

[1] https://docs.google.com/a/mozilla.com/document/d/1SgpqUXnOLCakx5HdDoLkv83Pt4iNUxeQMfV0iQ3fqBo/edit?usp=sharing

Comment 1

[Chris Pearce [:cpearce (Not reading bugmail)]]

Blake: Do you want to own removing or renewing and watching this telemetry? Neither Gerald or MattWoodrow are watching it, so unless you want to watch this, we should remove it.

Comment 2

[Blake Wu [:bwu][:blakewu]]

Chris, thanks for letting me know.
Kaku, is this telemetry useful to us? If yes, let's renew it.