python: properly propagate shutdown signal in weblogs

[florentinl]

Motivation

Re-enable this test: tests/test_resource_renaming.py::Test_Resource_Renaming_Stats_Aggregation_Keys::test_stats_aggregation_with_method_and_endpoint.

The main problem is that the weblog never receive the SIGTERM signal from Docker and get killed right away without having the possibility to flush their buffers on shutdown.

This PR ensures that the SIGTERM signal properly propagates to the python interpreters running the app and and can be caught by dd-trace-py.

Changes

• Use exec to replace the shell process with the actual ddtrace-run command.
• for uwsgi, updating + adding a bit more configuration was required to get the signal to properly propagate to the python interpreter
• re-enable the test

Workflow

1. ⚠️ Create your PR as draft ⚠️
2. Work on you PR until the CI passes
3. Mark it as ready for review
   • Test logic is modified? -> Get a review from RFC owner.
   • Framework is modified, or non obvious usage of it -> get a review from R&P team

🚀 Once your PR is reviewed and the CI green, you can merge it!

🛟 #apm-shared-testing 🛟

Reviewer checklist

• Anything but tests/ or manifests/ is modified ? I have the approval from R&P team
• A docker base image is modified?
  • the relevant build-XXX-image label is present
• A scenario is added, removed or renamed?
  • Get a review from R&P team

[github-actions]

CODEOWNERS have been resolved as:

manifests/python.yml                                                    @DataDog/apm-python @DataDog/asm-python
tests/test_resource_renaming.py                                         @DataDog/system-tests-core
utils/build/docker/python/django/app.sh                                 @DataDog/apm-python @DataDog/asm-python @DataDog/system-tests-core
utils/build/docker/python/fastapi/app.sh                                @DataDog/apm-python @DataDog/asm-python @DataDog/system-tests-core
utils/build/docker/python/flask/app.sh                                  @DataDog/apm-python @DataDog/asm-python @DataDog/system-tests-core
utils/build/docker/python/flask/requirements-uwsgi-poc.txt              @DataDog/apm-python @DataDog/asm-python @DataDog/system-tests-core
utils/build/docker/python/flask/uwsgi.ini                               @DataDog/apm-python @DataDog/asm-python @DataDog/system-tests-core
utils/build/docker/python/flask/uwsgi_app.sh                            @DataDog/apm-python @DataDog/asm-python @DataDog/system-tests-core
utils/build/docker/python/tornado/app.sh                                @DataDog/apm-python @DataDog/asm-python @DataDog/system-tests-core

[datadog-datadog-prod-us1-2]

✅ Tests

🎉 All green!

❄️ No new flaky tests detected
🧪 All tests passed

_{This comment will be updated automatically if new data arrives.
🔗 Commit SHA: c102ca1 | Docs | Datadog PR Page | Was this helpful? Give us feedback!}

[VianneyRuhlmann]

Nice, thx for fixing it

[chatgpt-codex-connector]

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit: c102ca18a4

ℹ️ About Codex in GitHub

Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you

• Open a pull request for review
• Mark a draft as ready
• Comment "@codex review".

If Codex has suggestions, it will comment; otherwise it will react with 👍.

Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".

[cbeauchesne]

For my knowledge, what are the diff between using exec and not, and why it makes the app more stable ?

[florentinl]

In the previous setup: the main process (pid 1) was bash (defined in the ENTRYPOINT of the upstream python images) and at startup bash runs our app.sh scripts. When docker wants to stop the container, it sends a SIGTERM (graceful shutdown) to the main process (bash) that does not forward signals to child processes and the app never has the occasion to run the flushing logic on graceful shutdown. docker then sends after a grace period a SIGKILL that kills the app at the kernel level.

When using exec, the launched process replaces the current process so ddtrace-run becomes the main process (pid 1) and docker sends its SIGTERM directly to ddtrace-run and we have the possibility to properly flush everything on shutdown.

If the flushing logic on shutdown works correctly (and it seems to be the case), that would mean that we can now set the library interface timeout back to 0 (from 5s currently)

[florentinl]

PHP images use dumb-init for the same purpose if I understand correctly, this is just a leaner version for when you have a single process

[cbeauchesne]

Thank you very much !