Closed Bug 970687 Opened 10 years ago Closed 10 years ago

Correlations tab in Socorro not loading

Categories

(Socorro :: General, task)

x86
macOS
task
Not set
normal

Tracking

(Not tracked)

RESOLVED FIXED

People

(Reporter: lizzard, Assigned: rhelmer)

Details

(Whiteboard: [QA+])

The correlations tab for this bug isn't loading anything:

https://crash-stats.mozilla.com/report/list?product=Firefox&range_value=7&range_unit=days&date=2014-02-10&signature=gfxDWriteFont%3A%3AComputeMetrics%28gfxFont%3A%3AAntialiasOption%29&version=Firefox%3A28.0b#tab-correlations

I kept the tab open for hours and tried a couple of different crash signatures for Firefox28 top crashers to no avail. Thanks for taking a look!
Hmm ok something is wrong with the query this is making:

https://crash-stats.mozilla.com/correlation?product=Firefox&version=28.0&platform=Windows%20NT&signature=gfxDWriteFont%3A%3AComputeMetrics%28gfxFont%3A%3AAntialiasOption%29&correlation_report_type=core-counts

That returns HTTP 400 with:
"""
version
    Select a valid choice. 28.0 is not one of the available choices.
"""

28.0b is a valid choice right now - however the "b" doesn't make it into the correlation scripts (it doesn't join against the product_versions table so doesn't have access to that info)

:peterbe - I think this is a bug, I am not sure if we should fix it in the Django view/model/mware service or in the correlation scripts. I am loathe to try to apply this fix to the scripts at this point, since they are more-or-less functional and we're hoping to replace them this quarter.

I think the right fix is to pass the "b" to the URL above, but then strip it off when the mware goes to fetch the file from crash-analysis. Any thoughts/preferences?
Assignee: nobody → rhelmer
Status: NEW → ASSIGNED
Seems like quite a straight forward bug in how the URL for the correlations partial gets constructed. It really should be 28.0b. 
I think I wrote most of the work around how partials are loaded. Let me know if you need help.
(In reply to Peter Bengtsson [:peterbe] from comment #2)
> Seems like quite a straight forward bug in how the URL for the correlations
> partial gets constructed. It really should be 28.0b. 
> I think I wrote most of the work around how partials are loaded. Let me know
> if you need help.

It's not very straightforward unfortunately - the results of the ReportList model (from /report/list mware service) do not include "b" (since those don't come in with the crashes) - this is used to pick the "crashiest" version to use for the correlation request. The correlation reports themselves have no concept of "beta" either, but that part is easier to fix.

I'm going to see build type (beta, release, etc) can be added to the /report/list service.
Commits pushed to master at https://github.com/mozilla/socorro

https://github.com/mozilla/socorro/commit/b9beaca817ec19931b5cb3f4fe141daf8937019f
fix bug 970687 - pass beta to correlations form, strip it out before
passing to service

https://github.com/mozilla/socorro/commit/b8de4e845b12cb14e2e7e81cbff529cf1bb89970
Merge pull request #1873 from rhelmer/bug970687-correlations-beta-not-working

fix bug 970687 - pass beta to correlations form, strip it out before
Status: ASSIGNED → RESOLVED
Closed: 10 years ago
Resolution: --- → FIXED
Target Milestone: --- → 74
There's an implicit bug here too, in that the loading indicators continue spinning even though the XHR failed. Filed bug 971215 to track that.
You need to log in before you can comment on or make changes to this bug.