Validate client_id / first_run_date fix in &-browser
Categories
(Toolkit :: Telemetry, task, P1)
Tracking
()
| Tracking | Status | |
|---|---|---|
| firefox68 | --- | affected |
People
(Reporter: Dexter, Assigned: chutten)
Details
(Whiteboard: [telemetry:mobilesdk:m8])
In this PR we fixed a bug that caused some clients to send pings that were not conformant to our schema, so they were dropped at ingestion, due to missing client_id and first_run_date.
This happened because old builds were installed, then not updated for a while, and finally updated to the latest version of &-browser, that contained a bunch of changes. This resulted in:
- &-browser knowing that it was not a "first run"
1a. so not generating new client_id/first_run_date - existing data living in a different storage file
- glean looking for things in the wrong place
We should check, at least, that:
- we see more pings than we had before (pings were getting dropped at ingestion);
- the error streams look less overloaded by "missing client_id" errors (see here)
|
Reporter | |
Updated•1 year ago
|
|
Assignee | |
Comment 1•1 year ago
|
||
Errors are less frequent: https://pipeline-cep.prod.mozaws.net/dashboard_output/graphs/analysis.moz_generic_error_monitor.org_mozilla_reference_browser.html
DAU counts are up, and the WAU trend has reversed from down to up: https://sql.telemetry.mozilla.org/queries/62314/source
Evidence is consistent with the theory that the client_id fix fixed things.
Description
•