consider adding telemetry for crash report submissions success/failure and unsubmitted crash
Categories
(Toolkit :: Crash Reporting, enhancement)
Tracking
()
| Tracking | Status | |
|---|---|---|
| firefox128 | --- | fixed |
People
(Reporter: willkg, Assigned: gerard-majax)
References
Details
Attachments
(1 file)
As we make changes to the crash reporter and to the collector, it would really help to have guardrail telemetry allowing us to see whether we've fixed things or made them worse.
For example, we periodically renew certificates and supporting Windows 7 users is tricky. It would be nice to know if we suddenly lost visibility on crash reports for Windows 7 users if we made a mistake renewing the certificate (wrong CA, etc).
For "failed to submit", the collector returns rejection codes:
https://socorro.readthedocs.io/en/latest/spec_crashreport.html#collector-response
Telemetry around "how many crash report submissions result in failure?" and "how often do we see which failure code?" would be really helpful.
Comment 1•2 years ago
|
||
We could use this to catch a number of problems we've experienced in the past including:
- Malformed crash reports which would be refused (so we'd never see them)
- Incomplete crash reports (same as above)
- Discrepancies between the number of reported crashes and the crash rate (to check if we somehow broke the crash submission machinery)
In the front-end we could listen to the crash-report-status event to send telemetry, or wire it up directly in the CrashSubmit machinery.
For full browser crashes we have to wait for bug 1759175 to be ready (which should be soon) then we'll be able to use that as it will have Glean support.
| Assignee | ||
Updated•1 year ago
|
| Assignee | ||
Updated•1 year ago
|
| Assignee | ||
Comment 2•1 year ago
|
||
We could use:
- https://mozilla.github.io/glean/book/reference/metrics/counter.html for submissions success, failures, unsubmitted
- https://mozilla.github.io/glean/book/reference/metrics/labeled_counters.html for colleting collector-response? https://socorro.readthedocs.io/en/latest/spec_crashreport.html#collector-response we only care about HTTP/400 right?
| Assignee | ||
Comment 3•1 year ago
|
||
Updated•1 year ago
|
Comment 5•1 year ago
|
||
| bugherder | ||
Description
•