prometheus-grafana

原文涵盖的主题

• Prometheus & Grafana
• Quick Start
• When to Use
• Best Practices
• Instrumentation
• Scraping & Storage
• Alerting
• Dashboards

Monitoring systems with Prometheus metrics and Grafana dashboards. Collecting time-series metrics from services Dashboards for latency, errors, and saturation Alerting on thresholds and SLO burn Kubernetes and infra monitoring Expose /metrics in Prometheus format Use counters (increasing), gauges, histograms Label metrics moderately (avoid cardinality explosion) Use client libraries for the language Set sane scrape intervals (15-60s) Use servicediscovery or static targets Limit retention and use TSDB best practices Configure recording rules for hot queries Define alerts with PromQL thresholds Use Alertmanager for routing and dedup Alert on symptoms (SLOs), not just metrics Set for, pending, and proper severity Build focused dashboards per service Use RED (Rate, Errors, Duration) and USE patterns Add panels with meaningful units Track SLOs with burn-rate panels 1. Instrument services to expose /metrics. 2. Configure Prometheus scraping and storage. 3. Add alerting rules and Alertmanager. 4. Connect Grafana as the data source. 5. Build dashboards with RED/USE patterns. 6. Track SLOs with burn-rate alerts. 7. Secure with auth and TLS. 8. Monitor the monitor: alert on Prometheus itself

这里只展示节选，请以完整原文和当前要求为准。