The graph is never generated for 'Beta' (and I think 'Release' as well). The 'Beta' result is always 'No results were found.'. E.g. https://crash-stats.mozilla.com/signature/?product=Firefox&signature=std%3A%3Alist%3CT%3E%3A%3Aclear%20|%20CDeviceChild%3CT%3E%3A%3A~CDeviceChild%3CT%3E#graph
Created attachment 8799423 [details] Screen Shot 2016-10-10 at 10.55.54 AM.png Clearly there are crashes under those release channels (see right-most column)
Adrian, There's no python error or anything. There simply is no records of 'beta' in crash_adu_by_build_signature. Here's for stage: breakpad=> SELECT breakpad-> channel, breakpad-> count(channel) as count breakpad-> FROM crash_adu_by_build_signature breakpad-> WHERE adu_date BETWEEN '2016-08-04' AND '2016-10-09' breakpad-> AND product_name = 'Firefox' breakpad-> -- AND channel = 'beta' breakpad-> AND signature = 'std::list<T>::clear | CDeviceChild<T>::~CDeviceChild<T>' breakpad-> group by channel breakpad-> ORDER BY count breakpad-> ; channel | count ---------+------- aurora | 132 nightly | 148 (2 rows) For that signature, there's plenty of crashes on beta. Just nothing in the ADU<->Signature mapping. See attached screenshot.
Created attachment 8799501 [details] Screen Shot 2016-10-10 at 3.56.27 PM.png Note the "Release channel" aggregate.
In fact, the are no rows in crash_adu_by_build_signature where channel='beta' and signature<>''. Sounds like a bug in how that stored procedure produces the data. Sigh.
To be honest, I'm actually surprised someone is using that graph... :) I believe we will want to rewrite it anyway, since it's using Postgres as a crash data source. Ideally it should use SuperSearch for crash data and marry that with ADI data from postgres.
We removed the Build Graph tab in https://bugzilla.mozilla.org/show_bug.cgi?id=1375431