OpenTelemetry

Raccoon accepts OTLP over HTTP, so any OpenTelemetry SDK can send to it without a collector in between. For Node.js and Next.js, raccoon-nodejs sets all of this up for you.

Configure the exporter#

Most SDKs read these standard variables. Use a token from Settings → Environments. The space after Bearer is written %20.

OTEL_EXPORTER_OTLP_ENDPOINT=https://raccoon.watch/api/otlp
OTEL_EXPORTER_OTLP_HEADERS=Authorization=Bearer%20<environment token>
OTEL_EXPORTER_OTLP_PROTOCOL=http/protobuf
OTEL_SERVICE_NAME=my-app
OTEL_METRICS_EXPORTER=none

• Raccoon takes OTLP over HTTP, as protobuf or JSON. It doesn't support gRPC, and some SDKs (.NET, and Python in some versions) default to gRPC, so set OTEL_EXPORTER_OTLP_PROTOCOL explicitly.
• The SDK adds /v1/traces and /v1/logs to the endpoint. If you configure each signal on its own, use https://raccoon.watch/api/otlp/v1/traces and https://raccoon.watch/api/otlp/v1/logs.
• Raccoon accepts metrics but doesn't store them, so turn the metrics exporter off.

For example, with Python:

pip install opentelemetry-distro opentelemetry-exporter-otlp
opentelemetry-bootstrap -a install
opentelemetry-instrument python app.py

What Raccoon reads#

Raccoon	OpenTelemetry
Requests	The outermost server span of each trace. The route comes from http.route.
Queries	Spans with db.system (or db.system.name) and a statement.
Outgoing Requests	Client HTTP spans.
Exceptions and Issues	exception span events. Stack traces from JavaScript, Python, Java, Go and Ruby are parsed into frames.
Logs	Log records, linked to their request through the trace id.
Users	user.id or enduser.id on the request span, with user.name, user.full_name or user.email.
Deploy markers	The service.version resource attribute.
Server	The host.name resource attribute, else service.instance.id, else service.name.

Jobs, commands, scheduled tasks, mail, notifications and cache are Laravel concepts, so those pages stay empty for OpenTelemetry apps.