Autoinstrumentation for opentelemetry ruby

[xuan-cao-swi]

Component(s)

auto-instrumentation

Is your feature request related to a problem? Please describe.

I'd like to open a feature request to include Ruby autoinstrumentation in the operator's autoinstrumentation.

The approach to autoinstrumentation should be similar to Node.js. Node.js uses NODE_OPTIONS to preload the OpenTelemetry library before loading or installing other libraries. Ruby can use RUBYOPT to preload the OpenTelemetry library.

The process of pod mutation is the same as other languages' autoinstrumentation (e.g., injectXXXSDK).

Autoinstrumentation script is waiting for review (PR)

Describe the solution you'd like

Current possible solution in PR 3756.

Describe alternatives you've considered

No response

Additional context

No response

[swiatekm]

For context, support for a given autoinstrumentation in the operator roughly involves two parts:

1. The autoinstrumentation image itself. The image needs to be built based on a Dockerfile and possibly some dependency manifest. The dependencies need to be updated, ideally as new versions of the instrumentation packages as released.

2. The operator logic for actually injecting the instrumentation into application Pods.

The operator maintainers are ready to take responsibility for point 2. Point 1 is more contentious, and we've had trouble maintaining the images, as we're not experts in the languages involved.

Here's what I think we need to decide before we accept a new auto-instrumentation:

Who is going to own the Dockerfile? I would strongly prefer if this was @open-telemetry/ruby-approvers.
How are we going to update the libraries in the aforementioned Dockerfile? Who is responsible for this?
How do we ensure the autoinstrumentation works as expected, including sending expected data?

CC @open-telemetry/operator-approvers