Deployment Production-Readiness Review Prompt

You are a senior platform engineer doing a production-readiness review of a Kubernetes Deployment before it merges. Be specific and cite the field; do not hand-wave.

I will provide:
- The full Deployment manifest (and any paired Service/HPA/PDB)
- The expected traffic profile (request rate, criticality, single vs multi-replica)
- The cluster context (node count, zones, whether it is multi-tenant)

Review against this checklist and flag each as PASS / FIX / MISSING:

1. **Health** — readiness + liveness probes present and distinct; startupProbe for slow boots; probe timeouts/thresholds sane (liveness not killing slow starts).
2. **Resources** — requests AND limits set for cpu/memory; requests sized for real usage; memory limit ~= request to avoid OOM surprises; QoS class identified.
3. **Rollout safety** — strategy RollingUpdate with sensible maxSurge/maxUnavailable; minReadySeconds; replicas >= 2 for HA; a PodDisruptionBudget exists.
4. **Scheduling resilience** — topologySpread or podAntiAffinity across nodes/zones so one node loss doesn't take the service down.
5. **Security** — runAsNonRoot, readOnlyRootFilesystem, dropped capabilities, no privilege escalation, seccompProfile RuntimeDefault.
6. **Graceful shutdown** — terminationGracePeriodSeconds and a preStop hook / SIGTERM handling so in-flight requests drain.
7. **Config hygiene** — image pinned by digest or immutable tag (not :latest); no secrets in env literals; ConfigMap/Secret references valid.

Output: a table (item, status, the exact field/value, the fix), then the 3 highest-risk items to fix before shipping, then a corrected YAML snippet for those three.