712030 - 9.0 Beta 6 Query data does not match

Status: RESOLVED INCOMPLETE

(Socorro :: General, task)

Description

[Marcia Knous [:marcia]]

When looking at query data for Beta 6 I noticed something odd:

This query shows 640 crashes for Js::Shape::finalize signature:

https://crash-stats.mozilla.com/query/query?product=Firefox&version=Firefox%3A9.0b6&platform=mac&range_value=1&range_unit=weeks&date=12%2F19%2F2011+09%3A14%3A02&query_search=signature&query_type=contains&query=&reason=&build_id=&process_type=any&hang_type=any&do_query=1

But I did a query for Mac specific crashes in B6 and it was showing a different number that was higher (although for some reason it is not now)

https://crash-stats.mozilla.com/query/query?product=Firefox&version=Firefox%3A9.0b6&platform=mac&range_value=1&range_unit=weeks&date=12%2F19%2F2011+09%3A19%3A44&query_search=signature&query_type=contains&query=&reason=&build_id=&process_type=any&hang_type=any&do_query=1

Comment 1

[Laura Thomson :laura]

If you reproduce this again, please note times of page loads.  These pages are cached, so new crashes may have come in between times, which is intentional behavior.

We can't do much about this without a test case, otherwise.  Reopen if you see it again.

Comment 2

[Robert Helmer [:rhelmer]]

(In reply to Marcia Knous [:marcia] from comment #0)
> When looking at query data for Beta 6 I noticed something odd:
> 
> This query shows 640 crashes for Js::Shape::finalize signature:
> 
> https://crash-stats.mozilla.com/query/
> query?product=Firefox&version=Firefox%3A9.
> 0b6&platform=mac&range_value=1&range_unit=weeks&date=12%2F19%2F2011+09%3A14%3
> A02&query_search=signature&query_type=contains&query=&reason=&build_id=&proce
> ss_type=any&hang_type=any&do_query=1
> 
> But I did a query for Mac specific crashes in B6 and it was showing a
> different number that was higher (although for some reason it is not now)
> 
> https://crash-stats.mozilla.com/query/
> query?product=Firefox&version=Firefox%3A9.
> 0b6&platform=mac&range_value=1&range_unit=weeks&date=12%2F19%2F2011+09%3A19%3
> A44&query_search=signature&query_type=contains&query=&reason=&build_id=&proce
> ss_type=any&hang_type=any&do_query=1

Actually looking more closely at this, the only difference between these two URLs is the time - the first one is 5 minutes before the other one and in that time 4 more crashes came in.

Here are two queries with the date specified as tomorrow (to make sure we get it real-time) and restricted to Mac:

https://crash-stats.mozilla.com/query/query?product=Firefox&version=Firefox%3A9.0b6&platform=mac&range_value=1&range_unit=weeks&date=12%2F20%2F2011&query_search=signature&query_type=contains&query=&reason=&build_id=&process_type=any&hang_type=any&do_query=1

Here it is for all platforms:

https://crash-stats.mozilla.com/query/query?product=Firefox&version=Firefox%3A9.0b6&range_value=1&range_unit=weeks&date=12%2F20%2F2011&query_search=signature&query_type=contains&query=&reason=&build_id=&process_type=any&hang_type=any&do_query=1

Right now the numbers match up for js::Shape::finalize as 675 for Mac on each (note there is one for linux).

Comment 3

[Robert Helmer [:rhelmer]]

(In reply to Laura Thomson :laura from comment #1)
> If you reproduce this again, please note times of page loads.  These pages
> are cached, so new crashes may have come in between times, which is
> intentional behavior.

I may have been wrong about the /query page being cached on the server-side, although it does appear that we set the "Expires" header to tell the browser to cache for 1 hour.

Marcia, does comment 2 help clear this up at all?