monitoring-patternslisted

# Monitoring Patterns ## When to load this skill - Python project with `from prometheus_client import` or `import prometheus_client` - User asks about Prometheus, metrics, counters, histograms, or observability - User is implementing health check endpoints - User asks about monitoring, alerting, or instrumenting code ## When not to load this skill - Infrastructure Prometheus configuration (scrape configs, alert rules, recording rules) - Grafana dashboard design - OpenTelemetry tracing (separate domain) --- ## Prometheus metric types | Type | When to use | Goes down? | | --------- | ------------------------------------------------------------- | ---------- | | Counter | Requests, errors, bytes processed - totals that only increase | No | | Gauge | Queue depth, active connections, memory usage - current state | Yes | | Histogram | Request duration, response size - distribution matters | No | | Summary | Like Histogram but computes quantiles client-side | No | Use **Histogram** over Summary by default. Summary quantiles are computed in the client and cannot be aggregated across instances. Histogram buckets are server-side and aggregatable. --- ## prometheus-client ```bash pip install prometheus-client ``` ### Counter Tracks cumulative values that only increase. ```python from prometheus_client import Counter REQUESTS = Counter( "http_requests_total"