1009990 - crashes not reported in health report

Status: RESOLVED WORKSFORME

(Firefox Health Report Graveyard :: Web: Health Report, defect, P2)

Description

[scott huxley]

User Agent: Mozilla/5.0 (Windows NT 6.3; WOW64; rv:32.0) Gecko/20100101 Firefox/32.0 (Beta/Release)
Build ID: 20140513030201

Steps to reproduce:

Goto > Nightly Health report.  


Actual results:

In my example see below:



    version32.0a1
    time open29751 min
    last crash2014-05-05
    bookmarks207

This Month

    time open8561 min
    crashes0



Expected results:

In health report the number of crashes reported should be equal to the number of instances your copy of Firefox Nightly has indeed crashed. But the accounting is not accurate.

Comment 1

[Robert Kaiser]

Benjamin, is this just a side-effect of our reworking of crash stats in FHR? If so, what bug will this be ultimately be fixed with, i.e. where to dupe to?

Comment 2

[Benjamin Smedberg]

No, this doesn't appear to be just a side effect, although that code probably also needs to be updated for the new crash format.

I have crashes in both formats in my report:

new day:
        "org.mozilla.crashes.crashes": {
          "_v": 2,
          "mainCrash": 2
        },
older day:
         "org.mozilla.crashes.crashes": {
          "_v": 1,
          "pending": 1
        }

And the interface still says 0 crashes.

Comment 3

[Robert Kaiser]

I guess we'll need updates in the FHR UI to account for the new format.

Comment 4

[Schalk Neethling [:espressive]]

Do we have a document/bug, or can someone explain what changed between the old format and the new? I can also log out the data I receive now and see what the code used to expect to determine the difference but, might be quicker and more accurate if it is explained by Benjamin perhaps?

Also, is this something that is different only in beta/aurora/nightly versions of Firefox? Also, is this a change only to desktop FHR? Thanks!

Comment 5

[Benjamin Smedberg]

It's always documented in http://mxr.mozilla.org/mozilla-central/source/services/healthreport/docs/dataformat.rst and it was implemented in a combination of bug 875562 and bug 983313.

Comment 7

[Benjamin Smedberg]

Schalk, this is a user-visibile problem that's probably pretty simple to fix. Can you prioritize it?

Comment 8

[Schalk Neethling [:espressive]]

Sure thing, let me take it.

Comment 9

[Schalk Neethling [:espressive]]

Hey Lonnen,

So, after speaking with Ben, I reckon it is best if bugs such as this one, and others related to FHR get reassigned to some on the web engineering team. If whomever takes this up has any questions, they most definitely are more than welcome to ping me.

Comment 10

[Schalk Neethling [:espressive]]

For ease of reference, here are two other bugs I am aware of:
https://bugzilla.mozilla.org/show_bug.cgi?id=885549
https://bugzilla.mozilla.org/show_bug.cgi?id=885040

Comment 11

[Schalk Neethling [:espressive]]

And here is another one:
https://bugzilla.mozilla.org/show_bug.cgi?id=888410

Comment 12

[Lonnen :lonnen]

For now just un-assign them. We don't have any spare capacity for FHR work right now.

Comment 13

[Benjamin Smedberg]

https://github.com/mozilla/fhr-jelly/pull/170

Using NI in lieu of a review request.

Comment 14

[Schalk Neethling [:espressive]]

r+ Thanks for taking this!

Comment 15

[Georg Fritzsche [:gfritzsche]]

Per bug 867123, comment 8 crashes seem to be displayed ok after the rewrite that switched this to Telemetry data.
Please open new bugs for any future issues.