I'd like to imagine that there is no crash in these versions... The issue is the same for 57.0b1 & 57.0b2 (see [1]). [1] https://crash-stats.mozilla.com/search/?release_channel=aurora&product=Firefox&date=%3E%3D2017-11-09T20%3A46%3A48.000Z&date=%3C2017-11-16T20%3A46%3A48.000Z&_sort=-date&_facets=version&_columns=date&_columns=signature&_columns=product&_columns=version&_columns=build_id&_columns=platform#facet-version

My bad, 58.0b2 never shipped but 58.0b1 did (57.0b1 & 57.0b2 shipped too).

58.0b3 only 93 crashes in the last 7 days: https://crash-stats.mozilla.com/search/?release_channel=aurora&version=58.0b3&product=Firefox&date=%3E%3D2017-11-09T15%3A46%3A48.000Z&date=%3C2017-11-16T15%3A46%3A48.000Z&_sort=-date&_facets=version&_columns=date&_columns=signature&_columns=product&_columns=version&_columns=build_id&_columns=platform#facet-version It looks like 58.0a1 has a lot of crashes: https://crash-stats.mozilla.com/search/?version=58.0a1&product=Firefox&date=%3E%3D2017-11-09T10%3A10%3A25.000Z&date=%3C2017-11-16T10%3A10%3A25.000Z&_sort=-date&_facets=version&_columns=date&_columns=signature&_columns=product&_columns=version&_columns=build_id&_columns=platform#facet-version but that's the nightly channel. Socorro has no ADI data for 58.0b1: https://crash-stats.mozilla.com/api/ADI/?end_date=2017-11-17&platforms=Mac%20OS%20X&platforms=Linux&platforms=Windows&platforms=Unknown&product=Firefox&start_date=2017-10-02&versions=58.0b1 but does have ADI data for 58.0b3: https://crash-stats.mozilla.com/api/ADI/?end_date=2017-11-17&platforms=Mac%20OS%20X&platforms=Linux&platforms=Windows&platforms=Unknown&product=Firefox&start_date=2017-10-02&versions=58.0b3 What does Telemetry say for usage of 58.0b1?

58.0b1 was devedition (aurora channel) only, 58.0b3 is the first build shipped on the beta channel, and socorro has no ADI for aurora AIUI. There are some crashes from the 58.0b1 build, except in crash-stats they show up as 58.0b0, e.g. https://crash-stats.mozilla.com/report/index/3233c77c-069a-49ca-af47-a3ab10171116 with a "release channel is aurora but no version data was found - added "b0" suffix to version number" note.

Looks like they're coming in with version "58.0", but since they're aurora channel, the processor adds "b0" to the end of that. Doing a supersearch shows this: https://crash-stats.mozilla.com/search/?release_channel=aurora&product=Firefox&version=58.0b0&date=%3E%3D2017-11-06T10%3A10%3A56.000Z&date=%3C2017-11-20T10%3A10%3A56.000Z&_sort=-date&_facets=signature&_columns=date&_columns=signature&_columns=product&_columns=version&_columns=build_id&_columns=platform#crash-reports Maybe that's where all the 58.0b1 crashes are? Why are these crashes coming in with version "58.0"?

I believe all builds off of mozilla-beta come with version 58.0. IIRC socorro has a scraper for archive.mozilla.org that maps buildid to version number for beta, maybe it doesn't do that for devedition?

Ok... This is making more sense. Socorro has an ftpscraper cron job that scrapes archive.mozilla.org for data. I think that populates the product_versions table. There's a processor rule that kicks in for crashes for the aurora and beta channels and does a lookup in the product_versions table to get the correct version number. I'll follow through on that. Maybe ftpscraper isn't scraping archive.mozilla.org and picking up aurora builds? Maybe the query Socorro uses isn't right somehow? Grabbing this to puzzle through today.

The "archive url has devedition, but the product in the build info is Firefox and the release channel is aurora with beta-style version numbers--and by the way, .0b1 is really the Firefox aurora build" thing breaks a number of invariants in Socorro. The consequence is that the entries we need in product_versions table aren't there, so that rule in the processor doesn't kick in to fix things. This is a big ball of spaghetti that I need to unravel and make sense of. We've got too many things up in the air to tackle this, too. Relatedly, I started working on replacing ftpscraper and friends with pulling the data from buildhub, but that work stalled out in favor of other things. I'll try to get that effort going again. Maybe it'll be up in time for 59. No promises! In the meantime, I'll look at writing a processor rule that fixes the data for specific builds by buildid. I can probably get that done by next week. Then we can reprocess crashes. Making this a P1 to do soon.

where 'fixes the data for specific builds by buildid' means it should be available in supersearch but likely will not work in other parts of socorro for a while

Commits pushed to master at https://github.com/mozilla-services/socorro https://github.com/mozilla-services/socorro/commit/fa71b0c4eb959443bb6cd93031560a75b270b3eb fixes bug 1418037 - add AuroraVersionFixitRule The AuroraVersionFixitRule checks the crash and stomps on the version if the crash is from Firefox with specific build ids. The build ids are for builds that are on the aurora channel for devedition product, but have a .0b1 beta-style version. We're fixing the crashes in the processor because the alternative was to work through the ball of spaghetti that is ftpscraper, the stored procedures, and related database table data to figure out why it's doing everything WRONG! The Socorro team therapist suggested we do a processor rule now and rewrite ftpscraper later and/or switch to buildhub. https://github.com/mozilla-services/socorro/commit/248ff49aa6c7a4050579ba29766ac0a57c3bd010 Merge pull request #4212 from willkg/1418037-aurora fixes bug 1418037 - add AuroraVersionFixitRule

This landed in master. We'll push it to -prod next week. After we push it, I'll set it up to reprocess all the 58.0b0 crashes.

We pushed to -prod yesterday and I reprocessed around 131k crashes, so I think we're all set now: https://crash-stats.mozilla.com/search/?build_id=20171103003834&product=Firefox&version=58.0b0&date=%3E%3D2017-10-26T19%3A19%3A16.000Z&date=%3C2017-11-26T00%3A19%3A16.000Z&page=1&_sort=-date&_facets=signature&_columns=date&_columns=signature&_columns=product&_columns=version&_columns=build_id&_columns=platform#crash-reports