Redpanda Connect: Debugging

Redpanda Connect (formerly Benthos) provides built-in tools for validating configs before deployment, tuning structured logs, exporting metrics to Prometheus or StatsD, shipping traces to OpenTelemetry, and probing pipeline health over HTTP. This skill covers the full debugging lifecycle: lint before you deploy, tune logging to see what is happening, scrape metrics to quantify it, and use the health endpoints to build reliable probes.

The HTTP server (default 0.0.0.0:4195) provides /ping (liveness), /ready (readiness — 200 only when both input and output are connected), and /metrics//stats (metrics). Use debug_endpoints: true to expose pprof.

Quickstart

# 1. Lint a config (exit 1 on any error)
rpk connect lint ./pipeline.yaml