Ignite Monitoring the right way
summary
Prioritizes monitoring work: get metrics first (agent + Prometheus/OpenMetrics) for CPU/memory and business metrics, then log aggregation, then simple alerting/visuals, and tracing last. Recommends JSON logs with UTC timestamps, log all application errors, never log PII, and avoid info-level logs in prod because they overflow systems. Treat alerts as real-impact only—drop “notifications.” Notes push vs pull is mostly irrelevant and points to OpenTracing/OpenCensus and Practical Monitoring as practical resources.
AI-written summary · about
