Instance States

Every instance has a state that reflects what it is currently doing. You can see the current state in the instance list and on the instance detail view. The state also governs which Actions are available at any given time.

Normal Lifecycle

In normal operation an instance is Healthy. Changing its configuration marks it Pending Update until you apply the change with Update; Update, Restart, and Recover each return it to Healthy when they finish, and Destroy is terminal.

stateDiagram-v2
    direction TB
    state "Pending Update" as PendingUpdate
    state "Building Image" as BuildingImage
    state update_choice <<choice>>
    [*] --> Creating: Create
    Creating --> Healthy
    Healthy --> PendingUpdate: on configuration change
    PendingUpdate --> update_choice: Update
    update_choice --> BuildingImage: requires a new image
    update_choice --> Updating
    BuildingImage --> Updating: build succeeds
    BuildingImage --> PendingUpdate: build fails
    Updating --> Healthy
    Healthy --> Restarting: Restart
    Restarting --> Healthy
    Healthy --> Recovering: Recover
    Recovering --> Healthy
    Healthy --> Destroying: Destroy
    Destroying --> [*]

Error Handling

If an operation fails, the instance enters Errored. From there you can correct the change and Update to retry, Recover to redeploy the last known-good image, or Destroy the instance.

stateDiagram-v2
    direction LR
    [*] --> Errored: an operation fails
    Errored --> Updating: Update
    Updating --> Healthy
    Errored --> Recovering: Recover
    Recovering --> Healthy
    Errored --> Destroying: Destroy
    Destroying --> [*]