This repo demonstrates how to use the OpenTelemetry span metrics connector to extract metrics data from spans. The span metrics connector aggregates Request Error and Duration OpenTelemetry metrics from span data. The spans and metric data are sent to New Relic.
- Node
- New Relic API key
npm install
In docker/collector-config.yml replace ${NEW_RELIC_API_KEY} with the API key from your New Relic account
Start the OpenTelemetry Collector with the following command
npm run docker:start
Start the OpenTelemetry Collector with the following command
npm run docker:startHead
Start the OpenTelemetry Collector with the following command
npm run docker:startTail
In a separate terminal window run the API locally using
npm run instrumentation:start
The application should start running on localhost:3000. You can make requests to the API using the following routes:
localhost:3000/api/products - Get all products
localhost:3000/api/products/:id - get product by ID
localhost:3000/api/products/search?query=capsule - search products by name or description (case-sensitive)
To shutdown the collector container
npm run docker:stop
Once you start generating requests you should find the service capsule-corp in the entity explorer of your New Relic account. Go to distributed traces and you'll see the requests populate. long with the traces if you head to the Metrics Explorer you'll see the metrics calls and duration. Once selected you can then facet by dimensions.
The head sampling demo makes use of the OpenTelemetry probabilisitc sampling processor. The demo is setup to get metrics from all incoming traces, but only collect 20% of traces.
The tail sampling demo makes use of the OpenTelemetry tail sampling processor. You will get metrics from all incoming traces, but only collect traces that generate errors.