nodejsagent provides a set of complementary instrumentation features for collecting relevant data to be processed by Hypertrace.

1. Update protobuf definitions
   1. Update the agent-config submodule.

      git submodule update --init --recursive
      

   2. Generate protobbuf files: ./src/instrumentation/config/generated.js and ./src/instrumentation/config.generated.d.ts.

      npm run generate_pb
      

2. Build.

   npm run build
   

You can then install the generated .tgz as a normal dependency.

Note: Using npm link doesnt solve the local testing problem as it will use the hypertrace package dev-dependencies during instrumentation instead of the targetted node app

• Run the example application

  1. cd ./examples/correlation
     

  2. Update config.yaml if needed.
  3. Install dependencies

     npm install
     

  4. Start application

     npm start
     

• Unit tests

  1. Start external services

     cd ./test/externalServices
     docker-compose up -d
     cd -
     

  2. npm run test
     

  3. Stop external services if not needed.

     cd ./test/externalServices
     docker-compose down
     cd -
     

  Tests should be added to the test directory, in a structure that matches that of the file you are attempting to test.

  Ideally all functionality is testable in locally runnable unit-tests(as opposed to in a docker container), primarily for ease of debugging.

To build the layer & upload it to an AWS account:

1. Build using the steps above.
2. Run ./build_layer.sh <REGION> where region is the region you want the layer to be available in
3. Add an environment variable to your lambda: AWS_LAMBDA_EXEC_WRAPPER=/opt/hypertrace-instrument
4. Configure reporting to a collector, you can set HT_REPORTING_ENDPOINT to a valid OTLP collector address. - If using with a collector layer, no need to specify an OTLP address as it will export to localhost:4317