creating install.json file related to apm single step instrumation

[schneideradam]

What does this PR do?

This PR makes the agent install ansible collection write an install.json file to the same directory as the Agent config file, datadog.yaml. The install.json file contains a random UUID, an installation timestamp, and the mechanism that was used to install the agent (the hardcoded string ansible_manual or ansible_single_step). The agent will send this information to the Datadog backend, so that we can gather telemetric data on how long users take to onboard after the initial installation of the agent.

Motivation

Jira: https://datadoghq.atlassian.net/browse/APMON-1070 . This is part of our broader work on improving the APM onboarding experience.

Additional Notes

Possible Drawbacks / Trade-offs

I am not copying these changes to the agent-dmg, iot-agent-deb or iot-agent-rpm directories, because we are initially focusing on improving the onboarding experience on Linux.

Describe how to test/QA your changes

Quick test

There isn't a quick test because of the nature of the ansible installer. Please follow the non-quick test below.

End to end test

To test, I first spun up a gcp RHEL instance and pulled the latest branch code. I installed ansible-core v2.11.12. I confirmed that the datadog_apm_instrumentation_enabled was set to "host".
From there I ran the playbook with the manual test listed in the command below:

ansible-playbook -i localhost, manual_tests/test_7_full.yml --connection=local

I checked in /etc/datadog-agent to ensure install.json existed and had the following values:
{"install_id": "88d5cfef-2f72-4b59-af8d-1e07b71b684b", "install_type": "ansible_single_step", "install_time": 1718116519}

I then changed the value for datadog_apm_instrumentation_enabled to an empty string "" and re-ran the command. I observed that the following install.json updated with the following:

{"install_id": "88d5cfef-2f72-4b59-af8d-1e07b71b684b", "install_type": "ansible_manual", "install_time": 1718116519}

I repeated all above steps in debian.

[yshapiro-57]

Is there any test plan that can be run to install the agent on Ansible with this code in place and verify that an install.json file gets generated?

[schneideradam]

I have verified this on both rhel and debian. I checked with both datadog_apm_instrumentation_enabled enabled, and disabled, and verified the correct behavior. Let me know if that is adequate. I can share screenshots if you'd like.

[yshapiro-57]

@schneideradam I trust you, you don't have to upload screenshots. I am just a fan of leaving test plans on PRs so that my successor knows how to maintain my code after I get promoted to VP and no longer have time to deal with this stuff.

Thank you very much again for helping me and @yahya-mouman on short notice with these changes. This was very nice of you.

[schneideradam]

@yshapiro-57 Do you mean you'd like to see some sort of CI unit test? Or instructions on how to test this PR? I'm happy to do whatever, I just usually don't worry about testing something like this when it should just fail silently so as not to get in the way of instrumenting their applications.

[yshapiro-57]

@schneideradam I was thinking something like this wiki or the end to end test plan on this PR; something that explains to a fresh person how to cause this code to be executed in a local environment.

Again, if you don't already have a writeup handy, and putting one together is time-consuming, then don't worry about it. I think it will be fine to document the testing steps on RHEL and Debian another time.

[schneideradam]

@spencergilbert Can we get this merged? Also, what is the process for deploying to ansible-galaxy?

[spencergilbert]

Followed up internally regarding release, expected publishing next week.

[Kyle-Verhoog]

@spencergilbert did the release go out?

cc @alexbryukhanova