Turn crash generation off by default for phone on stable channel

Bug #1494442 reported by Pat McGowan
22
This bug affects 3 people
Affects Status Importance Assigned to Milestone
Canonical System Image
Fix Released
High
Bill Filler
lxc-android-config (Ubuntu)
Fix Released
Undecided
Ken VanDine
ubuntu-system-settings (Ubuntu)
Fix Released
Undecided
Ken VanDine
unity8 (Ubuntu)
Fix Released
Undecided
Lukáš Tinkl

Bug Description

To avoid UI freezes and perceptions of slowness, we would like to turn this option off by default for consumers, but leave it on for developers and in the proposed channels.

Once bug #1437633 is fixed, we can put "manual" in the whoopsie.override file

Not sure how to do this conditionally just for stable

Tags: hotfix

Related branches

Changed in canonical-devices-system-image:
assignee: nobody → Pat McGowan (pat-mcgowan)
importance: Undecided → High
milestone: none → ww40-2015
status: New → Confirmed
tags: added: hotfix6
Revision history for this message
Jean-Baptiste Lallement (jibel) wrote :

I think it is not whoopsie that needs to be disabled but apport which is responsible of generating the crash file and is the source of the load on the device when a crash occurs. If the crash file is not uploaded then there is no need to generate it at all on a retail phone.

Revision history for this message
Oliver Grawert (ogra) wrote :

also the foundation team had some very strong opinions about turning off bug reporting by default, please involve them in the discussion.

Revision history for this message
Brian Murray (brian-murray) wrote :

jibel is correct and it is actually apport that would need to be disabled as it creates the crash files and creates the load on the device.

Revision history for this message
Iain Lane (laney) wrote :

Yes, I didn't understand what the request was for in the other bug - could have been about privacy concerns or something else.

summary: - Turn whoopsie reporting off by default for phone on stable channel
+ Turn crash greneration off by default for phone on stable channel
tags: added: hotfix
removed: hotfix6
summary: - Turn crash greneration off by default for phone on stable channel
+ Turn crash generation off by default for phone on stable channel
Revision history for this message
Pat McGowan (pat-mcgowan) wrote :

can we configure /etc/default/apport to set enable=1 only if system-image-cli reports the channel is proposed

Revision history for this message
Pat McGowan (pat-mcgowan) wrote :

 bug #1498133 should be addressed prior to disabling apport

Revision history for this message
Pat McGowan (pat-mcgowan) wrote :

The /etc/default/apport file should use the same setup as hostanme and timezone - being a link to /etc/writable
It can be removed from writeable_paths

see http://bazaar.launchpad.net/~ubuntu-core-dev/livecd-rootfs/trunk/view/head:/live-build/ubuntu-touch/hooks/03-etc-writable.chroot

Revision history for this message
Pat McGowan (pat-mcgowan) wrote :

Adding a settings task since the user option to send in reports needs to control this

Changed in canonical-devices-system-image:
assignee: Pat McGowan (pat-mcgowan) → Bill Filler (bfiller)
Revision history for this message
Ken VanDine (ken-vandine) wrote :

Added a unity8 task, the wizard will need to be updated. Currently there is a page in the wizard informing the user that their device is configured to automatically send crash reports. This will need to be updated to only be shown for devices that are configured to do so by default. There is a dbus service provided by whoopsie-preferences that provides this information.

interface = com.ubuntu.WhoopsiePreferences
property = AutomaticallyReportCrashes

Changed in unity8 (Ubuntu):
assignee: nobody → Lukáš Tinkl (lukas-kde)
Changed in unity8 (Ubuntu):
status: New → In Progress
Changed in ubuntu-system-settings (Ubuntu):
assignee: nobody → Ken VanDine (ken-vandine)
Changed in lxc-android-config (Ubuntu):
assignee: nobody → Ken VanDine (ken-vandine)
status: New → In Progress
Changed in ubuntu-system-settings (Ubuntu):
status: New → In Progress
Changed in unity8 (Ubuntu):
status: In Progress → Fix Committed
Changed in canonical-devices-system-image:
status: Confirmed → In Progress
Revision history for this message
Ted Gould (ted) wrote :

Seems a little extreme to disable all crash reporting. Why not disable it for Unity8, that seems to be the one that everyone is worried about. We have no other way to judge quality in the field, so it would be a real shame to loose the only feedback we have on real use-cases.

Revision history for this message
Ted Gould (ted) wrote :

Uhg, "lose"

Changed in canonical-devices-system-image:
status: In Progress → Fix Committed
Changed in lxc-android-config (Ubuntu):
status: In Progress → Fix Released
Changed in ubuntu-system-settings (Ubuntu):
status: In Progress → Fix Released
Revision history for this message
Matthew Paul Thomas (mpt) wrote :

Maybe the errors that testers experience are representative of the errors that customers experience. But even if that's true now, it won't necessarily be true all the time. There might be common errors that disproportionately/only affect customers using a language, time zone, keyboard, carrier, scope, or local app that our testers seldom use. Or errors that occur with the update process itself for a particular stable update. With error reporting turned off for stable, you won't know when this happens.

So, maybe a follow-up to this change could be to alter apport on stable to record just error statistics, not crash files. Then you wouldn't get a slowdown from dumping each crash, but you'd still be able to see when stable had an error rate higher than development channels for a particular package. When that happened, you could selectively turn on full error reporting for just that package to see what was going on.

Michał Sawicz (saviq)
Changed in canonical-devices-system-image:
status: Fix Committed → Fix Released
Changed in unity8 (Ubuntu):
status: Fix Committed → Fix Released
Revision history for this message
Jean-Baptiste Lallement (jibel) wrote :

Reverting to "committed" given that the image is not released yet

Changed in canonical-devices-system-image:
status: Fix Released → Fix Committed
Changed in canonical-devices-system-image:
status: Fix Committed → Fix Released
Revision history for this message
Brian Murray (brian-murray) wrote :

I was curious about how this changed the quantity of crash reports we receive from the stable channel. The following output shows the number of crashes per day from ubuntu-touch/stable/ubuntu armhf systems.

20151001: 53
20151002: 42
20151003: 38
20151004: 74
20151005: 38
20151006: 50
20151007: 50
20151008: 47
20151009: 51
20151010: 40
20151011: 51
20151012: 43
20151013: 53
20151014: 57
20151015: 57
20151016: 40
20151017: 37
20151018: 29
20151019: 35
20151020: 12
20151021: 14
20151022: 9
20151023: 13
20151024: 11
20151025: 5
20151026: 14
20151027: 18
20151028: 7
20151029: 18
20151030: 14
20151031: 14

That is a significant drop due to this change. The same query for crashes from ubuntu-touch/devel/ubuntu returns very few crashes.

20151002: 1
20151005: 1
20151006: 1
20151007: 1
20151008: 1
20151009: 1
20151014: 1
20151019: 1
20151115: 3

I think its really important we address the reason behind the UI freezes.

Revision history for this message
Brian Murray (brian-murray) wrote :

jibel suggested I check ubuntu-touch/rc-proposed/ubuntu:

20151001: 77
20151002: 89
20151003: 17
20151004: 16
20151005: 52
20151006: 56
20151007: 62
20151008: 43
20151009: 17
20151010: 7
20151011: 4
20151012: 19
20151013: 14
20151014: 39
20151015: 88
20151016: 71
20151017: 22
20151018: 11
20151019: 72
20151020: 54
20151021: 68
20151022: 47
20151023: 57
20151024: 7
20151025: 8
20151026: 90
20151027: 77
20151028: 72
20151029: 99
20151030: 73
20151031: 30

ubuntu-touch/rc-proposed/bq-aquaris.en:

20151001: 36
20151002: 45
20151003: 12
20151004: 17
20151005: 39
20151006: 20
20151007: 27
20151008: 9
20151009: 16
20151010: 4
20151011: 4
20151012: 8
20151013: 18
20151014: 26
20151015: 42
20151016: 56
20151017: 8
20151018: 13
20151019: 50
20151020: 35
20151021: 44
20151022: 45
20151023: 37
20151024: 14
20151025: 8
20151026: 78
20151027: 79
20151028: 88
20151029: 96
20151030: 73
20151031: 20

And meizu-en:

20151001: 68
20151002: 46
20151003: 19
20151004: 24
20151005: 47
20151006: 19
20151007: 13
20151008: 16
20151009: 24
20151010: 5
20151011: 4
20151012: 9
20151013: 16
20151014: 37
20151015: 50
20151016: 90
20151017: 22
20151018: 27
20151019: 25
20151020: 49
20151021: 76
20151022: 46
20151023: 38
20151024: 30
20151025: 16
20151026: 81
20151027: 58
20151028: 47
20151029: 36
20151030: 33
20151031: 19

To post a comment you must log in.
This report contains Public information  
Everyone can see this information.

Duplicates of this bug

Other bug subscribers

Remote bug watches

Bug watches keep track of this bug in other bug trackers.