How much does crashreporting care about hdd vs ssd, antivirus and firewall data (esp. for early startup)
Categories
(Data Science :: General, task)
Tracking
(Not tracked)
People
(Reporter: Gijs, Unassigned)
References
(Blocks 1 open bug)
Details
(I was told to file here in #telemetry on IRC)
Brief Description of the request
Information on the use of the telemetry environment used in crash reporting and telemetry, and which information in that environment is critical (or used at all).
Business purpose for this request
Collecting this environment information slows down Firefox startup significantly. We would like to take some information that is expensive to collect out of the (early) startup path. When we do that, the choice is for some version of:
- don't do anything and accept the serious performance implications (the perf team strongly feel this is not a feasible option)
- crashes before we manage to collect this information will not have any information for these fields
- we could potentially cache some environment information from the last Firefox run and submit that instead (antivirus/firewall/hdd/ssd information is, after all, unlikely to change)
- we could stop including this information in crash reports altogether if we do not feel it is useful for that purpose (ie we currently send the entire telemetry environment, but I expect some of that information is not in fact relevant for crashes).
Requested timelines for the request or how this fits into roadmaps or critical decisions
Work on startup and mainthread IO during startup is part of Q2 goals for the frontend perf team, so a quick response would be helpful.
|
||
Comment 1•5 years ago
|
||
See also bug 1516260, comment 3 ff for some context.
It's unclear to the Telemetry team which information is actually required for crash analysis etc.
|
Reporter | |
Comment 2•5 years ago
|
||
OK, it seems data science aren't the right stakeholders for crash reporting, but meanwhile I got feedback in bug 1363586 about this so we're all good here.
Description
•