The web deployment is schedulable but unreliable. Add realistic requests/limits and make traffic wait for the health endpoint.
Objectifs
- Request 100m CPU and 128Mi memory
- Limit 500m CPU and 256Mi memory
- Add readiness HTTP GET /healthz on port 8080
- Add startup probe with at least 30 seconds of failure budget
Étapes suggérées
- Resource fields are under the container.
- Probe httpGet is also under the container.
- failureThreshold × periodSeconds is the startup failure budget.
Indices
Voir la solution
# Edit the container with resources.requests/limits, readinessProbe and startupProbe, then apply and inspect rollout.