dmzoneill Rollout Restart
Restart a deployment and monitor rollout. Uses devops persona (k8s tools).
Inputs
| Input | Type | Default | Purpose |
|---|---|---|---|
deployment |
string | required | Deployment name |
namespace |
string | required | Kubernetes namespace |
environment |
string | stage | stage, production, ephemeral |
wait |
bool | true | Wait for rollout to complete |
Workflow
1. Load Persona
persona_load("devops")
2. Pre-Restart Check
kubectl_describe_deployment(deployment="{{ deployment }}", namespace="{{ namespace }}", environment="{{ environment }}")- Verify deployment exists (replicas > 0)
3. Restart
kubectl_rollout_restart(deployment="{{ deployment }}", namespace="{{ namespace }}", environment="{{ environment }}")
4. Monitor (if wait)
kubectl_rollout_status(deployment="{{ deployment }}", namespace="{{ namespace }}", environment="{{ environment }}")
5. Post-Restart
kubectl_get_pods(namespace="{{ namespace }}", environment="{{ environment }}")- Filter pods for this deployment; count Running/Pending/Failed
6. Optional: Verify Environment
skill_run("environment_overview", '{"namespace": "{{ namespace }}", "environment": "{{ environment }}"}')
7. Error Recovery
- "unauthorized" →
kube_login(cluster="stage" or "prod") - "deployment not found" →
kubectl_get_deployments()to list available
8. Memory
memory_session_log("Restarted deployment", "{{ deployment }} in {{ namespace }}, healthy={{ healthy }}")
Output
Report: namespace, environment, restart status, before state (replicas, image), rollout status, pod health (Running/Pending/Failed), and log commands if unhealthy.