feat(llmobs): support in-code config for llmobs

[lievan]

Support in-code configuration for LLMObs users, to enable LLMObs and specify the following configurations that currently require environment variable configuration.

• ml_app
• list of integrations to patch (will patch all LLMObs integrations by default)
• dd_llmobs_no_apm (turn off APM, telemetry, remote config, metrics)
• DD site, DD env, DD service (will override config/env vars)

from ddtrace.llmobs import LLMObs

LLMObs.enable(
    ml_app="comms/langchain", 
    integrations=["openai"],
    llmobs_agentless_enabled=True,
    # api_key =...
    # site=...
    # env=...
    # service=...
    # _tracer=None
)

Allowing in-code setup also improves the dev experience for people tracing experimental apps with LLMObs. It also abstracts away a long list of environment variables non-APM customers are required to set to turn off all APM related features.

This PR should not break any previous way of setting up the Python SDK (e.g. using env vars and ddtrace-run).

Arguments passed to enable() should take precedence over environment variables, with the exception of DD_LLMOBS_ENABLED.

This PR also does a couple minor things:

• If DD_LLMOBS_NO_APM env var is detected or configured through LLMObs.enable(), the OpenAI and LangChain integrations will disable submitting metrics unless the corresponding env vars DD_{OPENAI,LANGCHAIN}_METRICS_ENABLED is set to True.
• We also automatically disable both telemetry writer and remote config pollers if DD_LLMOBS_NO_APM is detected or configured through LLMObs.enable().
• We automatically patch the LLMObs integrations on LLMObs.enable().
• Removes all LLMObs.enable() references in individual integration patch code (openai, botocore, langchain)

Note:

• This change (only for LLMObs users) will override config.service, config.env if these are passed in to LLMObs.enable().
• If a user runs via ddtrace-run, they cannot use LLMObs.enable() to configure their settings.

Checklist

• Change(s) are motivated and described in the PR description
• Testing strategy is described if automated tests are not included in the PR
• Risks are described (performance impact, potential for breakage, maintainability)
• Change is maintainable (easy to change, telemetry, documentation)
• Library release note guidelines are followed or label changelog/no-changelog is set
• Documentation is included (in-code, generated user docs, public corp docs)
• Backport labels are set (if applicable)
• If this PR changes the public interface, I've notified @DataDog/apm-tees.

Reviewer Checklist

• Title is accurate
• All changes are related to the pull request's stated goal
• Description motivates each change
• Avoids breaking API changes
• Testing strategy adequately addresses listed risks
• Change is maintainable (easy to change, telemetry, documentation)
• Release note makes sense to a user of the library
• Author has acknowledged and discussed the performance implications of this PR as reported in the benchmarks PR comment
• Backport labels are set in a manner that is consistent with the release branch maintenance policy

[datadog-dd-trace-py-rkomorn]

Datadog Report

Branch report: evan.li/in-app-config
Commit report: edb3cd4
Test service: dd-trace-py

✅ 0 Failed, 405 Passed, 3511 Skipped, 20m 58.35s Total duration (1h 5m 38.25s time saved)

[sabrenner]

Some non-blocking comments, but one important one about just verifying that we're OK with prioritizing environment variables over config options. This is on me, as I had it backwards when we discussed it last week. Node, for example, does prioritize config options over environment variables. I think the rationale is config options could always be set by environment variables, ie:

LLMObs.enable(
  ml_app=os.environ["DD_LLMOBS_APP_NAME"],
  # etc...
)

So grabbing the config option first could be seen as more relevant (depends on the developer practices of the end user too I guess).

I do think that what we currently have (env var over config) could be fine, but only as long as we're consistent with it. Otherwise, I think swapping the order of precedence could be considered a breaking change.

It's just difficult to interpret intention 😅

[Yun-Kim]

Some comments

[Yun-Kim]

LGTM, great work!

[github-actions]

The backport to 2.9 failed:

The process '/usr/bin/git' failed with exit code 1

To backport manually, run these commands in your terminal:

# Fetch latest updates from GitHub
git fetch
# Create a new working tree
git worktree add .worktrees/backport-2.9 2.9
# Navigate to the new working tree
cd .worktrees/backport-2.9
# Create a new branch
git switch --create backport-9172-to-2.9
# Cherry-pick the merged commit of this pull request and resolve the conflicts
git cherry-pick -x --mainline 1 bf858f73768263ea7ed9743fff87d67d213afc11
# Push it to GitHub
git push --set-upstream origin backport-9172-to-2.9
# Go back to the original working tree
cd ../..
# Delete the working tree
git worktree remove .worktrees/backport-2.9

Then, create a pull request where the base branch is 2.9 and the compare/head branch is backport-9172-to-2.9.