Snappy update behavior after rollback

[Manik Taneja]

> On Dec 29, 2015, at 4:51 AM, Martin Albisetti <martin.albisetti at canonical.com> wrote:
> 
>> On Mon, Dec 28, 2015 at 10:41 PM, Manik Taneja <manik at canonical.com> wrote:
>> On my snappy kvm image, I was using the stable channel with ubuntu-core v9
>> and noticed that an update to v10 was available. The system failed the
>> update and as designed rolled back to v9 automatically. However, after the
>> system recovered at v9, the auto-update (aka autopilot) continued to update
>> the system at top of the hour each hour, and kept failing continuously. If
>> this state is reached, we should have built in logic to make sure that we
>> don't keep updating the system after a failed update or # of re-tries, which
>> can be over-ridden by the user (as needed).
> 
> In general, this is probably where health checks will come in. It'll
> mark a specific version as "faulty" and skip that version in future
> updates.
Thanks Martin. Do we happen to have a draft spec for me to review?

> 
> 
> -- 
> Martin