Getting started¶

The listener needs a UDP port reachable from the reporting endpoints and an OTLP metrics-and-logs destination. Start from the tracked example configuration:

workdir=$(mktemp -d)
cp config.example.yaml "$workdir/config.yaml"

Set otlp.endpoint to the base endpoint of a collector or backend, then retain only the sender addresses that are allowed to report. Sender names are stable metric values; they must be a small, explicit registry.

Run a local binary¶

Build and inspect the version:

make build
./bin/rfc6035-2otel --version

Start the binary with --config config.yaml after setting a reachable OTLP endpoint. The process binds 0.0.0.0:5060 by default. A clean shutdown on SIGINT or SIGTERM flushes its OTLP providers.

Container and Helm¶

The published image is ghcr.io/rknightion/rfc6035-2otel; use a release tag in production, mount configuration read-only, and expose UDP port 5060. The OCI chart is oci://ghcr.io/rknightion/charts/rfc6035-2otel. Its default Service is a UDP LoadBalancer and requests source-address preservation with externalTrafficPolicy: Local. Confirm that the target provider supports UDP load balancers and that this locality trade-off has enough ready endpoints. The chart README documents values, including sender configuration and secret handling for OTLP headers.

Liveness probe¶

The --healthcheck <listener-host>:5060 option sends a SIP OPTIONS probe to a running listener and expects its deliberate 405 Method Not Allowed response. It checks UDP reachability and the SIP response path; it does not prove that an OTLP backend received telemetry.