When the main workload exits, scoutd must report the final exit status back to the host, flush guest disk state, and power the microvm off cleanly. The host side runtime will later use this contract to decide whether the microvm succeeded, failed, or should be replaced.
The shutdown path must be deterministic and must behave correctly even after guest side failure.
Why
The guest lifecycle is not complete until the host can distinguish clean exits from crashes and the microvm shuts down in a controlled way.
In scope
- terminal exit status reporting
- final flush with sync
- deterministic guest power off
- child exit handling inside PID 1
When the main workload exits, scoutd must report the final exit status back to the host, flush guest disk state, and power the microvm off cleanly. The host side runtime will later use this contract to decide whether the microvm succeeded, failed, or should be replaced.
The shutdown path must be deterministic and must behave correctly even after guest side failure.
Why
The guest lifecycle is not complete until the host can distinguish clean exits from crashes and the microvm shuts down in a controlled way.
In scope