Fix: TF Agent host charm rebuilds COS scrape jobs based on supervisor config.

[gntzio]

Description

After deploying the Grafana Agent hosts in TEL, the /etc/grafana-agent.yaml config was not properly populated with scrape endpoints.

Following investigation, the suspicion is that since charm.py is instantiated on every Juju event, the in-memory self.scrape_endpoints variable is not consistently populated.

To address this, COSAgentProvider.scrape_configs are now sourced directly from the supervisord configuration.

Resolved issues

CERTTF-1038

Documentation

Charm unit tests.

Web service API changes

N/A

Tests

Charm unit tests.

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 77.21%. Comparing base (ebf71bd) to head (da8c9a7).
⚠️ Report is 2 commits behind head on main.

Additional details and impacted files

@@           Coverage Diff           @@
##             main    #1027   +/-   ##
=======================================
  Coverage   77.21%   77.21%           
=======================================
  Files         116      116           
  Lines       12065    12065           
  Branches      996      996           
=======================================
  Hits         9316     9316           
  Misses       2520     2520           
  Partials      229      229           

Flag	Coverage Δ		*Carryforward flag
agent	75.78% <ø> (ø)
cli	91.80% <ø> (ø)		Carriedforward from ebf71bd
device	63.34% <ø> (ø)		Carriedforward from ebf71bd
server	86.92% <ø> (ø)		Carriedforward from ebf71bd

*This pull request uses carry forward flags. Click here to find out more.

Components	Coverage Δ
Agent	75.78% <ø> (ø)
CLI	91.80% <ø> (ø)
Common	∅ <ø> (∅)
Device Connectors	63.34% <ø> (ø)
Server	86.92% <ø> (ø)

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.
• 📦 JS Bundle Analysis: Save yourself from yourself by tracking and limiting bundle sizes in JS merges.

[rene-oromtz]

LGTM, as far as I remember, Grafana Agent required the relation with Prometheus to actually configure the scrape jobs.

Due to initial TEL deployment without Prometheus, that relation data may never got populated from our charm side... I think the key is actually the refresh_events addition which also makes sense because when we add new agents, we should also be refreshing the scrape_jobs with the new agents config.

[gntzio]

@ajzobro, thanks for pointing out potential improvements.
As far as I can see, they are (mostly) not related to the changes in this PR itself.
I would appreciate it, if these could be addressed in a separate "improvements" PR.

[ajzobro]

Please address the issues that can easily be addressed within the scope of the code that was introduced here, including tests for the failure conditions.

CERTTF-1041 now eixsts as a strategy for handling (kicking along) the technical debt this time.

[ajzobro]

If we write the tests for the error cases we will clearly see the issues in the code.

[ajzobro]

I guess I'm not getting any more tests, huh?

[ajzobro]

Agree with creation of 1041 for improved testing and fixing the relied-upon-but-out-of-scope functions.

[rene-oromtz]

Thanks @ajzobro! Agree that there is lots to improve. At least now you already identify what needs to be done it should be a bit faster to implement in 1041