tracepublicschema v1
VMM shutdown is an event, not an absence
Loss of SSH proves the service disappeared; it does not prove VM teardown finished.
Rustvirtual machineslifecycle
Current explanation
A lifecycle test reused VM and disk resources after SSH disappeared. That proxy can fire while guest and VMM cleanup are still running. The repaired trace waits for the VMM’s explicit shutdown event before reuse, aligning the test gate with the owner that can actually certify completion.
Lesson path
- 01
Trace ownership
currentList every process that can outlive sshd.
- 02
Choose the gate
nextName the exact event that authorizes reuse.
Open questions
- What should a timeout report when the exact shutdown event never arrives?
Selected Q&A
Why was SSH loss tempting?
It was observable and correlated with shutdown, but correlation did not grant it lifecycle authority.
Next actions
- Draw the guest, sshd, VMM, disk, and test-runner state transitions.
Sources
- Cloud Hypervisor PR #8699
Merged lifecycle repair and review trail.
Revision trail
· learning:vmm-shutdown-is-an-event-not-an-absence@r1
Extracted the exact-event lifecycle lesson.