OpenTelemetry JS Instrumentation Digma

Digma facilitates continuous feedback by gleaning valuable insights from your code and displaying them in the IDE as you code.

This OpenTelemetry instrumentation package for Node.js helps Digma analyze your code by adding a number of OTEL resource attributes to the spans.

Prerequisites

• Node version: 8 or above.

Installing the module

npm install @digma/otel-js-instrumentation

Usage

Setup

• Instrumenting your OpenTelemetry resource
• Adding instrumentation for specific server frameworks
• Exporting trace data to Digma
• Fine tuning and ehhancements

Instrumenting your OpenTelemetry resource

Digma needs to add a few more attributes to your OTEL Resource. To update your OTEL setup, simply use the provided digmaAttributes function as seen below:

const { digmaAttributes } = require('@digma/otel-js-instrumentation');

const sdk = new opentelemetry.NodeSDK({
  resource: new Resource({
    [SemanticResourceAttributes.SERVICE_NAME]: 'my-service',
    ...digmaAttributes({
      rootPath: __dirname,
    }),
  }),
  spanProcessor: new BatchSpanProcessor(exporter),
  instrumentations: [getNodeAutoInstrumentations()],
});

Adding instrumentation for specific server frameworks

Digma can also generate specifc insights based on the service framework you're using. To do that, we can add a simple middleware that will save the contexual information needed to map the tracing data to the underlying code in the Span attributes.

Follow the steps in the below links to add Digma's middlware, based on your server framework:

• https://github.com/digma-ai/digma-instrumentation-express

For example, here is how you would use Digma's middlware with Express along with the standard OTEL middlware:

const { digmaRouteHandler } = require('@digma/instrumentation-express');

app = express();
app.use(digmaRouteHandler);
app.use('/users', users); // some router example

Exporting trace data to Digma

First, you need to have a Digma backend up and running. You can follow the instructions in the Digma project repository to quickly get set up using Docker.

You can use a standard OTLP exporter for local deployments:

const otlpExporter = new OTLPTraceExporter({
  url: 'http://localhost:5050',
});

Alternative, if you're already using a collector component you can simply modify its configuration file:

exporters:
...
otlp/digma:
    endpoint: "localhost:5050"
    tls:
      insecure: true
service:
  pipelines:
    traces:
      exporters: [otlp/digma, ...]

In both cases, set the endpoint value to the URL of the Digma backend.

That's it! You should be good to go.

Fine tuning and ehhancements

Digma allows you to set additional attributes while setting up the OpenTelemetry Resource to allow better observability visualization for commits, deployment environments, and more. All of these are optional, but can help provide more context to the collected traces:

| Options | Input Type | Attribute Key | Description | Default | | --- | --- | --- | --- | --- | | rootPath | string | [code.package.path, code.package.name]| rooPath describes the absolute path of the package.json file | None | digmaEnvironment | string | digma.environment | The Environment describes where the running process is deployed. (e.g production, staging, ci) | See the note on the Digma environment below. | commitId | string | scm.commit.id | The specific commit identifier of the running code. | | otherPackages | [] string | code.package.others | Specify additional satellite or infra packages to track | |

A note on the Digma environment

To support different configurations and deployment strategies, Digma provides a few ways to set the Digma environment. This list describes the order of precedence:

1. If set, the DEPLOYMENT_ENV environment variable has the highest priority and always wins.
2. If the environment variable is not set, Digma will use the digmaEnvironment attribute value passed to the digmaAttributes function.
3. If no deployment environment is provided, we'll assume this is a local deployment environment and mark it using the local hostname. It will be visible only to this machine.