Health and dependencies

Readiness and liveness

Each probe declares its own type: http, tcp, or command.

healthcheck:
  readiness:
    type: http
    url: http://127.0.0.1:3000/ready
    interval: 2s
    timeout: 1s
    failure_threshold: 10
  liveness:
    type: tcp
    port: 3000
    initial_delay: 10s
    interval: 15s
    timeout: 2s
    failure_threshold: 3

Readiness gates dependent startup. Liveness describes a running service's ongoing health. Lifecycle status for a detached resource is separate: a resource may exist while readiness is still failing.

Dependency conditions

services:
  api:
    command: npm run dev
    depends_on: [database, migrate]
    dependency_conditions:
      database:
        condition: process_healthy
      migrate:
        condition: process_completed_successfully

Supported conditions:

Condition	Satisfied when
process_started	The process started, or detached status observed running
process_healthy	Readiness passed; lifecycle status alone is insufficient
process_completed	The command exited
process_completed_successfully	The command exited with a successful code
process_log_ready	ready_log_line matched captured output

Starting a target includes its transitive dependencies. Stopping a target includes transitive dependents in reverse order. Shift+S deliberately bypasses graph expansion and operates on the selected targets only.

Recovery

Process-supervised services can configure restart behavior:

availability:
  restart: on_failure
  backoff: 2s
  max_restarts: 5

Detached status changes do not trigger automatic restart. An external resource may have been intentionally stopped outside Kranz, so observation remains read-only until the user explicitly starts it.