1579522 - Don't send the "modules" ping within the first 30min of a profile's first session

Status: RESOLVED FIXED

(Toolkit :: Telemetry, task, P1)

Description

[Chris H-C :chutten]

Conversation starts here: https://bugzilla.mozilla.org/show_bug.cgi?id=1576948#c5

In summary, the "modules" ping is registered to a weekly cadence. The first tick in that cadence happens about 10-15min into the first session of a new profile. It doesn't need to happen at all that first week, let alone that first session. (see bug 1576948 comment 9)

So let's delay it somehow for some length of time. I can think of three options in decreasing pleasantness to implement:

1. Don't register the timer at all in that first session (by checking TelemetryReportingPolicy's isFirstRun. Might have to rewire it slightly to be a static pref getter because the value isn't correct until session restore's complete.)
2. Modify the update timer manager such that the first tick is configurable. (Specifically make this part optional.)
3. Store some state in ModulesPing to ignore timer ticks that happen to early. (Probably would involve adding a pref)

Comment 1

[Chris H-C :chutten]

Attachment: Bug 1579522 - Allow skipping first update timer interval r=kmag

The UpdateTimerManager has an undocumented behaviour for firing its listeners
very soon after first registration (about 10min into the first session).

Let's document that behaviour, and make it optional.

Comment 2

[Chris H-C :chutten]

Attachment: Bug 1579522 - Don't trigger a "modules" ping in the first session r=janerik,aklotz

Depends on D46292

Comment 3

[Pulsebot]

Pushed by [e-mail]:
https://hg.mozilla.org/integration/autoland/rev/fb1c3bf44fb1
Allow skipping first update timer interval r=kmag
https://hg.mozilla.org/integration/autoland/rev/229c49ee94df
Don't trigger a "modules" ping in the first session r=janerik,aklotz

Comment 4

[Noemi Erli[:noemi_erli]]

Backed out 2 changesets (Bug 1579522) for xpcshell failures in test_ModulesPing.js CLOSED TREE

Push with failure: https://treeherder.mozilla.org/#/jobs?repo=autoland&resultStatus=testfailed%2Cbusted%2Cexception&revision=229c49ee94dfffd39f37a750bb2524587570b2e4&searchStr=xpcshell&selectedJob=268034849

Failure log: https://treeherder.mozilla.org/logviewer.html#/jobs?job_id=268034849&repo=autoland&lineNumber=4882

Backout: https://hg.mozilla.org/integration/autoland/rev/ed3e58e3fc243a6abe5b11d62d0acdb92b8c8974

Comment 5

[Chris H-C :chutten]

Well that's odd. It appears to be attempting a network connection? That... doesn't make sense. And to au5.mozilla.org? The host that serves updates?

Not sure what this has to do with... ohhhhhh... maybe by futzing with the update timer manager I caused the other registered timers to bork somehow? Clearly I need to add more tests to update timer manager.

Comment 6

[BugBot [:suhaib / :marco/ :calixte]]

There are some r+ patches which didn't land and no activity in this bug for 2 weeks.
:chutten, could you have a look please?
For more information, please visit auto_nag documentation.

Comment 7

[Chris H-C :chutten]

Thanks auto_nag. I've been busy, and the r+ patches were backed out. I'm working on it now (oh why am I talking to a bot)

Comment 8

[Pulsebot]

Pushed by [e-mail]:
https://hg.mozilla.org/integration/autoland/rev/a69f8d7d2a53
Allow skipping first update timer interval r=kmag
https://hg.mozilla.org/integration/autoland/rev/fdc947cae3e5
Don't trigger a "modules" ping in the first session r=janerik,aklotz

Comment 9

[Razvan Maries]

https://hg.mozilla.org/mozilla-central/rev/a69f8d7d2a53
https://hg.mozilla.org/mozilla-central/rev/fdc947cae3e5