Bug 1295934 added a "StartupCrash" annotation to crash reports. This bug is about supporting/taking advantage of it in Socorro. This includes: - Add it to the Details tab in crash-stats. Putting it in the "crash circumstances" section just before "Process Type" makes sense. It might be worth omitting it if the value is 0, not sure. - Whitelist it. I think this is done by adding it RawCrash.API_WHITELIST. Does this also require modifying the SuperSearch database? The possible values for this annotation are 0 and 1. - Use it instead of the existing heuristic which, AIUI, is triggered when > 50% of crashes in a signature have an uptime of 0. This is the most significant change, because the previous changes just add information to crash reports, but this one changes how an existing thing is done. (Because the new annotation applies to individual crashes, not just groups of crashes, we could apply the little rocket icon to individual crashes as well, as is done with the AMD icon.) Anything else I've missed?

And perhaps some tests, e.g. in socorro/unittest/external/es/base.py.

If we do this, I'd like to have it prior to Socorro -- it could be useful in the ping as well as the report. (I say this, but I don't yet know if that's possible.)

It's just a straight annotation, so it's definitely available to the code that's generating the ping. I don't think gsvelto's patches currently pull any other info out of the annotations, but it should be straightforward to do so.

I did put "StartupCrash" in kCrashEventAnnotations[], which is "A whitelist of crash annotations which do not contain sensitive data and are saved in the crash record and sent with Firefox Health Report". So I don't know if it's already being included in the ping. (But note that my understanding of pings, FHR and telemetry is weak, so I might have that wrong.)

Just FYI, currently we determine if a signature is a startup crash by looking if 50%+ of its crashes have an uptime <60 seconds (the code is misleading, because it looks at the the first interval called "0", which actually contains all document where uptime < 60).

(In reply to Ted Mielczarek [:ted.mielczarek] from comment #3) > It's just a straight annotation, so it's definitely available to the code > that's generating the ping. I don't think gsvelto's patches currently pull > any other info out of the annotations, but it should be straightforward to > do so. Currently I'm only adding stacks to the existing crash ping, and possibly the crash ID too. As for kCrashEventAnnotations it's used to filter what ends up in the event file so if you add an entry it will end up in the crash ping automatically AFAIK.

I've opened bug 1308474 for adding the field to SuperSearch and bug 1308476 for using the annotation instead of the current heuristic, and morphed this bug into just adding the field to the `Details` tab of the Report Index page.

Commit pushed to master at https://github.com/mozilla/socorro https://github.com/mozilla/socorro/commit/b4a795bb9bb02d745205d4b12c1670fcf0fec98c Fixes bug 1297966 - Show startup crash field in report index page (#3508) * Show the StartupCrash annotation in the report index page * Show a nicer 'True' instead of '1' * Display StartupCrash=True only when it is present and it is '1' * Define a jinja helper to transform a booleanish string to a boolean and use it for the StartupCrash annotation * Show the StartupCrash annotation also when it is false

I've verified this on stage: https://crash-stats.allizom.org/report/index/7f156e79-e4f4-451e-a68a-d992f2161007 https://crash-stats.allizom.org/report/index/c461963d-bf94-4603-b6e5-609e42161005 https://crash-stats.allizom.org/report/index/dfbc9aca-3eba-4325-b65c-3746f2161007 Yay!