Check and resume an incomplete host OS update (supervisor not updated)

[pdcastro]

For some reason (power outage, impatient end user, something else), a device rebooted before the host OS update got to the point of PATCH'ing the API for the supervisor version and, as a result, the device found itself with an updated balena OS version but the supervisor was kept at the previous version.

The goal is for this situation to be somehow detected and remedied on the next device boot. Like, say: "incomplete update detected, resuming update..."

[jellyfish-bot]

[pdcastro] This issue has attached support thread https://jel.ly.fish/12afcf7c-7731-447e-a140-c99dd20da454

[xginn8]

It may make sense to trigger the supervisor update earlier in the process, perhaps even immediately after pulling the new host app. I think it's a better situation generally (given a modern-looking OS, not counting pre-MC devices) to have an updated supervisor rather than a trailing supervisor. cc @CameronDiver

[jellyfish-bot]

[lucianbuzzo] This issue has attached support thread https://jel.ly.fish/07247f52-020a-4352-a46b-d9a1c49bee70

[CameronDiver]

@xginn8 I agree, as long as no rollback will occur.

[xginn8]

@CameronDiver do you mean supervisor rollback, or host OS rollback?

[CameronDiver]

Sorry I meant supervisor rollback, then remembered that doesn't happen. Ignore me.