646884 - No crash stats for "Firefox for mobile"

Status: RESOLVED WORKSFORME

(Socorro :: General, task)

Description

[Scoobidiver (away)]

Since the release of Firefox for mobile, there is no longer ADU in Fennec.
The crash signatures of Firefox for mobile are also not in Firefox's crash stats.

Comment 1

[Laura Thomson :laura]

ADUs come from a Metrics push.  ccing Daniel for info.

Comment 2

[Stephen Donner [:stephend] Not actively reading bugmail]

(Just FYI, WebQA is working on getting our test suite up and running in production, too, but it might a little time -- it would have caught this, I think.)

Comment 3

[Daniel Einspanjer [:dre] [:deinspanjer]]

We checked the ADU side of things and there is steadily increasing ADU for Fennec 4.0 in the Socorro database just as I expect.

I don't know what is going on here, but I don't see it being a problem with the ADU itself.

Comment 4

[Scoobidiver (away)]

The ADU is now fixed on March 31th.
But the 4.0 crash number seems to be incredibly low (ratio of 10) wrt. the 4.0 RC one (same tendency as desktop Firefox).

Firefox 4.0 for mobile:
https://crash-stats.mozilla.com/daily?form_selection=by_version&p=Fennec&v[]=4.0&throttle[]=100.00&v[]=&throttle[]=100.00&v[]=&throttle[]=100.00&v[]=&throttle[]=10.00&hang_type=any&os[]=Windows&os[]=Mac&os[]=Linux&date_start=2011-03-17&date_end=2011-03-31&submit=Generate

Firefox 4.0:
https://crash-stats.mozilla.com/daily?form_selection=by_version&p=Firefox&v[]=4.0&throttle[]=100.00&v[]=&throttle[]=100.00&v[]=&throttle[]=100.00&v[]=&throttle[]=10.00&hang_type=any&os[]=Windows&os[]=Mac&os[]=Linux&date_start=2011-03-17&date_end=2011-03-31&submit=Generate

Comment 5

[Daniel Einspanjer [:dre] [:deinspanjer]]

Is the throttle supposed to be 100% for both of those queries?

Comment 6

[Laura Thomson :laura]

Scoobidiver, a couple of things:

First, the ADU lags a day or so behind actual usage, since it depends on blocklist pings.  That would account for the ADU issue.

Second, Firefox 4.0 went from processing 100% of crashes to processing 10% at midnight on Monday.  Fennec was also switched over at the same time (this particular throttle rule depends purely on version).  We should have communicated this better.

In general we process 10% of crashes for Firefox and 100% for nightlies betas, RCs, Thunderbird, etc as their volume is much lower.  

Are you requesting that we process 100% of Mobile crashes for 4.0?  What's the expected ADU increase over time?

Comment 7

[Scoobidiver (away)]

> What's the expected ADU increase over time?
I don't know. There are 4 billions mobile phones in the world. If I suppose that there is only 1% that is compatible with Firefox (ARMv7 + Android), there are 40 millions could-be users (ratio of 10 with the desktop version). Moreover, the ramp-up may take time.

> Are you requesting that we process 100% of Mobile crashes for 4.0?
Mobile Firefox is not very crashy (ratio of 4 with desktop) with few users until now, so I would say yes.

Comment 8

[(not currently active) Ted Mielczarek]

We probably shouldn't throttle Mobile crashes, I think the crash volume there is so low that it's more harmful than useful. Mobile Firefox crashes should still be differentiable, since the product name is "Fennec" and not "Firefox".

Comment 9

[Laura Thomson :laura]

Yes, they are differentiable.  I'm going to close this out and open a bug for the throttle change.