Affects Version/s: None
Fix Version/s: None
The current acceptance criteria for
MEN-1187 are out of date, and a new plan for handling spontaneous reboots are required.
- Shall all scripts be rerun when a spontaneous reboot occurs?
- What Is going to guarantee critical functionality like network connectivity?
After some discussion these are the solutions that Kristian Amlie and I are pondering:
- Have Reboot leave be the state entered after a spontaneous-reboot, and thus have this state setup all the functionality required by
the other states.
- Implement a separate spontaneous-reboot state to handle it.
- Have each state (and transition) be atomic, meaning that each state/transition will need to have sentinel-scripts that guarantee the functionality required by that state. This means that
if a spontaneous-reboot occurs, the entire state/transition will be rerun.
Currently we are leaning towards the third choice, as it is the cleanest, with the least amount of added complexity.