Consider experimenting with early-shutdown pings
Categories
(Toolkit :: Telemetry, task, P1)
Tracking
()
Tracking | Status | |
---|---|---|
firefox112 | --- | fixed |
People
(Reporter: chutten, Assigned: travis_)
References
(Blocks 1 open bug)
Details
(Whiteboard: [telemetry:fog:m?])
Attachments
(1 file)
(deleted),
text/x-phabricator-request
|
Details |
If there are events recorded after the last inactivity in an app session, Glean sends them on the next app session's startup in an "events" ping with reason "startup" (as per docs). But maybe these events need not have this extra latency?
With Nimbus we can experiment with changing FOG's shutdown behaviour. Specifically, if we add in InitializeFOG
something like
RunOnShutdown([&] {
if (NimbusFeatures:GetBool("glean"_ns, "final_inactive"_ns, false)) {
fog_set_user_inactive();
}
}, ShutdownPhase::XPCOMWillShutdown);
(( You'd need to put glean::handle_client_inactive()
out on firefox_on_glean
's API surface as well as adding the named Nimbus Feature on Glean ))
And, if you want to get extra spicy, we've also been interested in perhaps sending a final "metrics" ping at the same time. We'd need to expand fog_submit_ping
to accept a reason
plus all the other stuff... but it could be need to fiddle with.
Assignee | ||
Updated•2 years ago
|
Assignee | ||
Comment 1•2 years ago
|
||
Comment 3•2 years ago
|
||
bugherder |
Description
•