feat(bedrock): support metrics for bedrock

[jinsongo]

• I have added tests that cover my changes.
• If adding a new instrumentation or changing an existing one, I've added screenshots from some observability platform showing the change.
• PR name follows conventional commits format: feat(instrumentation): ... or fix(instrumentation): ....
• (If applicable) I have updated the documentation accordingly.

[gitguardian]

️✅ There are no secrets present in this pull request anymore.

If these secrets were true positive and are still valid, we highly recommend you to revoke them.
Once a secret has been leaked into a git repository, you should consider it compromised, even if it was deleted immediately.
Find here more information about risks.

---

^{_{🦉 GitGuardian detects secrets in your source code to help developers and security teams secure the modern development process. You are seeing this because you or someone else with access to this repository has authorized GitGuardian to scan your pull request.}}

[jinsongo]

I already know the root cause why I always keep getting the error botocore.errorfactory.AccessDeniedException: An error occurred (AccessDeniedException) when calling the InvokeModel operation: You don't have access to the model with the specified model ID.. Thanks.

[gyliu513]

Is this required?

[jinsongo]

It's an open issue. I think it's better to use a common name, for example llm.completions.exceptions for all ai system to count exceptions. If yes, I would like to use another PR to fix all related code.

[nirga]

@jinsongo if we're doing this here I'd try to use the new genAI conventions - https://opentelemetry.io/docs/specs/semconv/gen-ai/gen-ai-metrics
I think they're already released so you can even use them directly from the original otel semconv package

[nirga]

(in this case - it means exceptions are counted in the duration metric)

[jinsongo]

@nirga I cannot find an appropriate metric name for the exceptions counter of Bedrock. As you know, version 0.4.1 of opentelemetry-semantic-conventions-ai does not include LLM_BEDROCK_COMPLETIONS_EXCEPTIONS although it's already included in this current PR commits, which causes the build to fail when I use it in the Bedrock instrumentation code. Actually, that's the same approach for OpenAI, Authropic and Watsonx.

[nirga]

@jinsongo this should be in the standard opentelemetry-semantic-conventions package. And you should count exceptions on the duration metric as specified in the spec.

[gyliu513]

@jinsongo If this parameter is not required, I think we can ignore this for now, and we can fix this in future as a consolidation for semantic convention.

[jinsongo]

@gyliu513 I have removed LLM_BEDROCK_COMPLETIONS_EXCEPTIONS from semconv_ai, and added a TODO comment about fixing in future as a consolidation for semantic convention.

[nirga]

nice @jinsongo! a couple of styling comments and we're ready to go

[gyliu513]

/lgtm

@nirga can you help review? Thanks

[jinsongo]

@nirga Do you have more comments? Thanks.