Child payloads may be missing in pings

RESOLVED FIXED

Status

()

Toolkit
Telemetry
P1
normal
RESOLVED FIXED
2 years ago
2 years ago

People

(Reporter: gfritzsche, Assigned: gfritzsche)

Tracking

Trunk
Points:
1

Firefox Tracking Flags

(e10s?, firefox48 affected)

Details

(Whiteboard: [measurement:client])

(Assignee)

Description

2 years ago
Bill reported seeing missing child payloads on Nightly Telemetry data for pings with e10s enabled.
(Assignee)

Updated

2 years ago
Priority: -- → P3
(Assignee)

Comment 1

2 years ago
I'm checking on this.
Assignee: nobody → gfritzsche
Points: --- → 1
Priority: P3 → P1
(Assignee)

Comment 2

2 years ago
Breakdown, using Aurora for presumably less noisy data:
https://gist.github.com/georgf/a97f774c2e2fec40edb16657e872032c

For the pings with missing child payloads:
* >60% of them had content crashes in the session
* Most of the remaining pings have short session lengths (<60s/80s or so)

Given that, i don't see a big problem here.
We could investigate the missing child payloads in short sessions more closely as part of the proper child payload design/collection work, but i don't think there is anything urgent or critical here.

Roberto, what do you think?
Flags: needinfo?(rvitillo)
Do we have a bug on file for not getting telemetry for crashed content processes? It seems like the sort of thing we should be sending up continuously, like session restore data.
Also, thanks for investigating.
(Assignee)

Comment 5

2 years ago
(In reply to Bill McCloskey (:billm) from comment #3)
> Do we have a bug on file for not getting telemetry for crashed content
> processes? It seems like the sort of thing we should be sending up
> continuously, like session restore data.

That would be one approach, but we should also consider whether that data matters etc.
I would suggest we approach this as part of the e10s child payload design meeting later this week, i added you as optional there.

Updated

2 years ago
tracking-e10s: --- → ?
(Assignee)

Updated

2 years ago
Summary: Child payloads often missing in pings → Child payloads may be missing in pings
(In reply to Bill McCloskey (:billm) from comment #3)
> Do we have a bug on file for not getting telemetry for crashed content
> processes? It seems like the sort of thing we should be sending up
> continuously, like session restore data.

I agree with Bill.
Flags: needinfo?(rvitillo)
(Assignee)

Comment 7

2 years ago
Closing this bug then.
I added the continuous sending to the preliminary design notes, we can go file bugs once we have a design.
Status: NEW → RESOLVED
Last Resolved: 2 years ago
Resolution: --- → FIXED
You need to log in before you can comment on or make changes to this bug.