Fix #2678 don't add loadmodule when from config

[remicollet]

See #2678

Track in moduleLoadQueueEntry where the module was loaded from

• config (moduleEnqueueLoadModule)*
• command (module load or loaded)

Only add loadmodule line when not from config to avoid duplicate line.

[codecov]

Codecov Report

❌ Patch coverage is 0% with 12 lines in your changes missing coverage. Please review.
✅ Project coverage is 72.17%. Comparing base (0646749) to head (b40ab88).

Files with missing lines	Patch %	Lines
src/module.c	0.00%	8 Missing ⚠️
src/config.c	0.00%	4 Missing ⚠️

Additional details and impacted files

@@             Coverage Diff              @@
##           unstable    #2689      +/-   ##
============================================
- Coverage     72.18%   72.17%   -0.01%     
============================================
  Files           128      128              
  Lines         71037    71041       +4     
============================================
- Hits          51277    51275       -2     
- Misses        19760    19766       +6     

Files with missing lines	Coverage Δ
src/module.h	0.00% <ø> (ø)
src/config.c	78.64% <0.00%> (-0.10%)	⬇️
src/module.c	9.77% <0.00%> (-0.01%)	⬇️

... and 12 files with indirect coverage changes

🚀 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.

[remicollet]

Notice: this fixes the loadmodule case

but can be considered as a workaround, the main issue is that the included configuration files are not managed during the rewrite process, so they can generate duplication.

[remicollet]

I have slightly reduced the PR by protecting only loadmodule from include files.
So for main configuration file, the behavior is unchanged

[zuiderkwast]

I think this is a bugfix and I think we should backport it. I can't see how it's a breaking change, or even a behavior change for any working configuration.

[zuiderkwast]

@remicollet Please fix the DCO issue. Essentially, commits need to be done with git commit -s. See the DCO failing check details page for information.

[remicollet]

@remicollet Please fix the DCO issue. Essentially, commits need to be done with git commit -s. See the DCO failing check details page for information.

Smashed and signed

[zuiderkwast]

Great, but author and signer emails need to match.

[remicollet]

Great, but author and signer emails need to match.

fixed

[madolson]

Core team meeting discussion:

1. We need to also validate the behavior for normal module configs when the module is unloaded and a user does CONFIG REWRITE as we might have the same issue with the server not starting because the module config will still be present.
2. We think CONFIG REWRITE doesn't make that much sense for production workloads, we should probably consider deprecating it or something to avoid people using it in the future, or more properly fix the behavior with CONFIG REWRITE and includes.
3. We think in the fullness of time we would like to add a way for operators to disable CONFIG REWRITE to prevent users from accidentally breaking applications. We will figure out the proper way to do that for 9.1 (if it's backwards compatible) or 10 (if it's behavior change).