Merge lp:~vorlon/lxc-android-config/apport-job-cleanup into lp:lxc-android-config
| Status: | Needs review |
|---|---|
| Proposed branch: | lp:~vorlon/lxc-android-config/apport-job-cleanup |
| Merge into: | lp:lxc-android-config |
| Diff against target: |
161 lines (+23/-101) 5 files modified
debian/changelog (+8/-0) debian/lxc-android-config.maintscript (+2/-0) etc/init/apport-config.conf (+13/-10) etc/init/apport-reconfig.conf (+0/-33) etc/init/apport.override (+0/-58) |
| To merge this branch: | bzr merge lp:~vorlon/lxc-android-config/apport-job-cleanup |
| Related bugs: |
| Reviewer | Review Type | Date Requested | Status |
|---|---|---|---|
| Ken VanDine | 2016-02-04 | Needs Fixing on 2016-02-17 | |
|
Review via email:
|
|||
This proposal supersedes a proposal from 2015-10-15.
Commit Message
Rework and simplify the apport overrides. Instead of adding two new upstart
jobs and an override for apport.conf, we can achieve everything we need to
with a single new apport job.
This leaves outstanding the issue that switching a device between channels
from the commandline will not cause the config to be correctly switched.
Description of the Change
Rework and simplify the apport overrides. Instead of adding two new upstart
jobs and an override for apport.conf, we can achieve everything we need to
with a single new apport job.
This leaves outstanding the issue that switching a device between channels
from the commandline will not cause the config to be correctly switched.
Verified to work correctly on a stock Ubuntu 14.04 desktop system; will need
validating against the phone images also.
| Ken VanDine (ken-vandine) wrote : | # |
QA failed this during landing. It seems to do the right thing at boot, but when toggling it via system-settings it no longer sets the core_pattern. It must not be starting and stopping the apport job on FILE=/var/
Not sure why though... but it does regress
| Brian Murray (brian-murray) wrote : | # |
It looks like there is an EVENT missing for the file watch, could that be the cause of the problem?
| Ken VanDine (ken-vandine) wrote : | # |
> It looks like there is an EVENT missing for the file watch, could that be the
> cause of the problem?
Oh, yeah I bet that's it, good catch. It needs the start on EVENT=create and stop on EVENT=delete.
| Steve Langasek (vorlon) wrote : | # |
On Fri, Feb 26, 2016 at 01:56:32PM -0000, Ken VanDine wrote:
> > It looks like there is an EVENT missing for the file watch, could that be the
> > cause of the problem?
> Oh, yeah I bet that's it, good catch. It needs the start on EVENT=create
> and stop on EVENT=delete.
FWIW I hadn't replied yet because I haven't had time to investigate for
sure, but I don't think this is the case. The EVENT= env var declaration
makes the match more specific; a match which does not include the EVENT=
should match a 'file' event with /any/ value of EVENT (or none at all).
Unmerged revisions
- 14. By Steve Langasek on 2016-02-04
-
Merge from trunk
- 13. By Steve Langasek on 2015-10-14
-
Further refinement: no apport override needed at all, do everything in the apport-config helper job
- 12. By Steve Langasek on 2015-10-14
-
rm -f is always polite
- 11. By Steve Langasek on 2015-10-14
-
Further reduction of the apport override
- 10. By Steve Langasek on 2015-10-14
-
Simplify apport handling, to avoid duplication of the main apport.conf
where avoidable and to simplify the event start model (avoiding possible
deadlocks).

This no longer merges cleanly, please update the branch