Skip to main content

Observability for Ignition

This pillar covers how to see what an Ignition gateway is doing in production: its metrics, logs, and traces, and the stack that collects and visualizes them. The patterns here are deployment-agnostic. They apply whether the gateway runs under Docker Compose, on Kubernetes, or on a bare-metal install, with the cloud-specific wiring called out where it differs.

What to monitor​

A useful picture of a running gateway combines a few signal families:

  • RED metrics (rate, errors, duration) for gateway and Perspective request traffic.
  • JVM health such as heap, garbage collection, and thread counts, since the gateway is a long-lived JVM.
  • Per-database connection-pool and query metrics for each configured database connection.
  • Perspective session and component activity.
  • MQTT / data throughput where the gateway moves tag data over MQTT or other transports.

How the pillar is organized​

The flagship guide wires the gateway's OpenTelemetry export, the agnostic foundation every other piece builds on:

  • Gateway Telemetry: attaches the OTel Java agent to the gateway process and configures it to push metrics, traces, and logs over OTLP. Start here.
  • Metrics and Log Stack: the Grafana LGTM collector and storage stack for Docker Compose, and the kube-prometheus-stack alternative for Kubernetes.
  • Log Pipeline: the Alloy pipeline that processes, filters, and routes gateway log records to Loki, with rationale for each filter rule.

The OTel Properties Reference lists every agent property used in the guides with a one-line description.

Further guides covering traces and profiling (Tempo, Pyroscope), external exporters, and custom Ignition metrics land in upcoming rounds. Short one-screen operational tips will be collected under a Tasks sub-section as the pillar grows, following the same Concepts / Tasks / Tutorials / Reference convention the rest of the site uses.

Scope​

For core-product concepts, this pillar links out rather than re-explaining: see docs.ia.io for the gateway internals and charts.ia.io for Helm chart value semantics. The guides here own the integration and operational glue: how to wire, compose, and operate the observability stack around a gateway.