| name | reliability-checklist |
| description | Production readiness verification for services and infrastructure. Use before promoting a service to production, after significant infrastructure changes, or during periodic production readiness reviews. |
| license | MIT |
Reliability Checklist
Verify production readiness of a service or infrastructure change.
When to Use
The production-readiness gate before promoting a service — a pass/fail audit
of operational readiness: resource limits, health/readiness probes, fault
tolerance, graceful degradation, right-sizing. It does not instrument the
service (building the logs/metrics/alerts is observability-setup) and does
not cover network-boundary correctness (that's
distributed-systems-checklist). This is the operational go/no-go, not the
telemetry or the wire protocol.
Checklist
- Resource limits: CPU and memory limits set on all containers/VMs. No unbounded resource consumption.
- Health checks: Liveness and readiness probes configured. Restart policy defined.
- Fault tolerance: Single points of failure identified and mitigated (replication, failover, load balancing).
- Graceful degradation: Service behaves predictably under partial failure (upstream down, DB slow).
- Right-sizing: Resources allocated based on actual usage data, not estimates. Review after first week of production load.
Completion
- All 5 checklist items verified with evidence (command output, config snippet)
- Findings reported: pass/fail per item with remediation for failures
Anti-Patterns
- Rubber-stamping — checking boxes without verifying actual state
- Skipping right-sizing — over-provisioning wastes resources, under-provisioning causes outages
- One-time check — reliability must be re-verified after significant changes
