834086 - Server backend for Android App Not Responding (ANR) reporting

Status: RESOLVED WONTFIX

(Mozilla Metrics :: Metrics Data Ping, defect)

Description

[(inactive) Jim Chen [:jchen] [:darchons]]

Pending reviews, we want to introduce automatic reporting of Fennec hangs (App Not Responding / ANR). This is the Fennec-equivalent of the chrome hang feature on Desktop.

The reporting process involves taking function call stacks and logs on a hung Android device and sending those data to Metrics through telemetry. Fennec will use the standard telemetry ping JSON format, with some exceptions:

1) Due to limitation in the Java front-end, some fields will not be available in the telemetry ping. For "simpleMeasurements", only "uptime" will be available. Basic fields in "info" will be available, but some of the optional fields in "info" will not be available.

2) Because call stacks are preprocessed on Android, instead of "chromeHangs" field, the Android telemetry ping will have an "androidANR" field containing stacks in text form. Therefore, no further symbol processing is needed.

3) The Android telemetry ping will also have an "androidLogcat" field containing Fennec logs in text form.

Therefore, for basic functionality, the server backend has to extract androidANR and androidLogcat fields from telemetry pings and save them to a protected location (similar to what's being done now for the desktop chrome hang feature).

Comment 1

[(inactive) Jim Chen [:jchen] [:darchons]]

Attachment: Sample ANR ping

Comment 2

[(inactive) Jim Chen [:jchen] [:darchons]]

Attachment: Sample ANR ping payload

Comment 3

[(inactive) Jim Chen [:jchen] [:darchons]]

Attachment: Sample ANR text

Comment 4

[(inactive) Jim Chen [:jchen] [:darchons]]

Attachment: Sample logcat text

Comment 5

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

This bug doesn't belong in Metrics.  I'm not sure if it belongs in Fennec or Socorro though so I can't move it myself.

Comment 6

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

I was mistaken.  Will discuss with team and get a more accurate follow-up.

Comment 7

[(inactive) Jim Chen [:jchen] [:darchons]]

Any update, Daniel?

Comment 8

[Mark Goodwin [:mgoodwin]]

(In reply to Jim Chen [:jchen :nchen] from comment #7)
> Any update, Daniel?

Do you want to include review of this part in the secreview (bug 833991) or should I pick this up with metrics separately (I'm assuming it's more a privacy thing anyway if the server changes are trivial).

Comment 9

[(inactive) Jim Chen [:jchen] [:darchons]]

(In reply to Mark Goodwin [:mgoodwin] from comment #8)
> (In reply to Jim Chen [:jchen :nchen] from comment #7)
> > Any update, Daniel?
> 
> Do you want to include review of this part in the secreview (bug 833991) or
> should I pick this up with metrics separately (I'm assuming it's more a
> privacy thing anyway if the server changes are trivial).

Hm, I don't know if metrics has their own procedure... The changes should be trivial though, because I believe much of the code already exists to process the telemetry pings and save them to protected storage.

Comment 10

[(inactive) Jim Chen [:jchen] [:darchons]]

Not relevant anymore because we have http://telemetry.mozilla.org/hang/anr now