@open-telemetry/javascript-approvers could you have a look? 🙏

Thank you @pavolloffay i fixed the build since getResourceDetectorsFromEnv is exported as getResourceDetectors

Actually, I think since metrics auto-configuration is now implemented, you may not even need the autoconfigure.ts file anymore. It just duplicates what's happening in @opentelemetry/auto-instrumentations-node/register now (the only difference being that the implementation from the upstream package does not throw when a non-existing exporter is selected)

Hi @pichlermarc, are you suggesting to remove autoinstrumentation.ts here and update the code in nodejs.go to something like:

nodeRequireArgument = " --require /otel-auto-instrumentation-nodejs/node_modules/@opentelemetry/auto-instrumentations-node/build/src/register.js"

Would this approach work when referenced like this? I haven’t tested it yet.

Thank you

PS @swiatekm, what’s your opinion on this?

Hi @pichlermarc, are you suggesting to remove autoinstrumentation.ts here and update the code in nodejs.go to something like:

nodeRequireArgument = " --require /otel-auto-instrumentation-nodejs/node_modules/@opentelemetry/auto-instrumentations-node/build/src/register.js"

Would this approach work when referenced like this? I haven’t tested it yet.

Thank you

PS @swiatekm, what’s your opinion on this?
It makes sense to me

Hi @pichlermarc, are you suggesting to remove autoinstrumentation.ts here and update the code in nodejs.go to something like:

nodeRequireArgument = " --require /otel-auto-instrumentation-nodejs/node_modules/@opentelemetry/auto-instrumentations-node/build/src/register.js"

Something along these lines yes. I don't think you necessarily need to reference the full path - you may be able to just ref the entrypoint:

nodeRequireArgument = " --require /otel-auto-instrumentation-nodejs/node_modules/@opentelemetry/auto-instrumentations-node/register"

@pavolloffay @pichlermarc could you re-review this (I assume) final state of the change?

@ElfoLiNk have you tested this with a real Node application? Our tests only check if the instrumented application starts, but not if it's actually emitting the right telemetry, so this should be verified manually with this kind of substantial change.

@pavolloffay @pichlermarc could you re-review this (I assume) final state of the change?

@ElfoLiNk have you tested this with a real Node application? Our tests only check if the instrumented application starts, but not if it's actually emitting the right telemetry, so this should be verified manually with this kind of substantial change.

I tested today, and everything seems to be working:

OTEL_LOGS_EXPORTER is empty. Using default otlp exporter.
Unsupported OTLP metrics protocol: "undefined". Using http/protobuf.
OpenTelemetry automatic instrumentation started successfully

I was testing it with this configuration:

- name: OTEL_TRACES_EXPORTER
  value: "otlp"
- name: OTEL_METRICS_EXPORTER
   value: "otlp"
- name: OTEL_NODE_RESOURCE_DETECTORS
   value: "env,os,process,container"

The exporter endpoint was set to port 4317.

The only issue I encountered was with the metrics exporter:

Error: PeriodicExportingMetricReader: metrics export failed (error Error: Parse Error: Expected HTTP/)

To resolve this, I explicitly added OTEL_EXPORTER_OTLP_PROTOCOL to the instrumentation configuration, which resolved the issue.

Additionally, both the operator and instrumentation Docker images need to be updated simultaneously. If you update only the instrumentation, you encounter the following error:

Error: Cannot find module '/otel-auto-instrumentation-nodejs/autoinstrumentation.js'