Operations¶

Guides¶

Quick Start — Get the stack running in 5 minutes
Replay Mode — Validate config changes against historical data
Monitoring — Metrics, dashboards, health checks
Troubleshooting — Common issues and solutions

Operational Model¶

graph LR
    A[Tune config] --> B[Replay test]
    B --> C{Results OK?}
    C -->|No| A
    C -->|Yes| D[Apply to running stack]
    D --> E[Monitor metrics]
    E --> F{Noise?}
    F -->|Yes| A
    F -->|No| G[Stable]

Current Deployment¶

Environment	Mode	Stack
Local (docker-compose)	Dry-run	Controller + 3 Workers + Redis + ML
Cluster (planned)	Dry-run → Live	K8s Deployment + IRSA + ArgoCD

Key Operational Metrics¶

Metric	Healthy range	Alert if
`staffops_ad_controller_cycle_duration_seconds`	< 15s	p99 > 30s
`staffops_ad_worker_queries_total{status="error"}`	< 5%	> 10%
`staffops_ad_ml_calls_total{status="error"}`	< 5%	> 10%
`staffops_ad_detection_anomalies_total`	Varies	Sudden 10x spike
`staffops_ad_alert_dedup_hits_total`	> 50% of fires	0% (dedup broken)