feat(aws_cloudwatch_logs sink): allow setting type of log class to cr…

[PriceHiller]

Summary

Copied from the body of commit:

Problem: Prior to this commit it was not possible to specify the log group's class type. The method prior always created a Standard type.

---

Solution: Allow specifying the log group class type via a new field, group_class which takes over the create_missing_group.

deprecation: create_missing_group

Initial Issue Report: #22008

Closes #22008

This PR allows the specification of the log class to choose instead of defaulting to only the Standard log class type.

I'm not sure the approach I've taken is ideal, and I'm unsure as to the full deprecation procedure in code for the create_missing_group field.

If the approach taken here with mirroring the LogClassGroupDef is not idiomatic/poor, I'm open to suggestions (and for anything else, of course) 🙂.

Change Type

• Bug fix
• New feature
• Non-functional (chore, refactoring, docs)
• Performance

Is this a breaking change?

• Yes
• No

How did you test this PR?

• By running: cargo test --no-default-features --features "sinks-console,sinks-aws_cloudwatch_logs,aws-cloudwatch-logs-integration-tests" -- --skip 'sinks::aws_cloudwatch_logs::integration_tests' 'sinks::aws_cloudwatch_logs'
• NOTE: I was unable to run the integration tests for the cloudwatch logs. There doesn't appear to be a entry for them in the Makefile.

Does this PR include user facing changes?

• Yes. Please add a changelog fragment based on our guidelines.
• No. A maintainer will apply the "no-changelog" label to this PR.

Checklist

• Please read our Vector contributor resources.
• If this PR introduces changes Vector dependencies (modifies Cargo.lock), please
  run dd-rust-license-tool write to regenerate the license inventory and commit the changes (if any). More details here.

References

• Closes: AWS CloudWatch Logs - create infrequent access type with create_missing_group #22008

[PriceHiller]

Ah, just realized I missed a changelog entry 😅, will add in a moment.

[PriceHiller]

Added two changelog entries in the latest force push.

[PriceHiller]

Latest force push caught a potential regression, realized I accidentally broke support for the old behavior supported by create_missing_group.

Now if create_missing_group is enabled, we set the LogGroupClass to Some(LogGroupClass::Standard) which reflects the previous behavior.

[jszwedko]

Thanks for this contribution @PriceHiller !

I think it isn't immediately apparent that specifying group_class will cause a group to be created. What would you think of leaving in create_missing_group without deprecation and have that control whether a group is created. group_class could then default to standard and only be used if create_missing_group is true. I realize it is a bit of redundancy, but I think the options will be more discoverable.

[PriceHiller]

Thanks for this contribution @PriceHiller !

I think it isn't immediately apparent that specifying group_class will cause a group to be created. What would you think of leaving in create_missing_group without deprecation and have that control whether a group is created. group_class could then default to standard and only be used if create_missing_group is true. I realize it is a bit of redundancy, but I think the options will be more discoverable.

I think you're right, what you've proposed is probably a better approach. I can modify this PR tomorrow (hopefully) to support that form of functionality instead.

Will ping when done, for now I'll convert this PR to a draft (unless you'd prefer a new PR and me to close this).

[jszwedko]

Thanks @PriceHiller ! This looks good to me now.

[PriceHiller]

Huh, interesting seeing those failures in the CI, I guess I get to state the most dangerous words in the world: "it works on my machine™".

I hate to be this much of a pain in the ass for less than a hundred line change, I just realized I didn't include the updated Cargo.lock 🤦.

Will add in a moment, and modify PR items as necessary.

[PriceHiller]

@jszwedko I do apologize for all the back and forth, just updated the Cargo.lock and the aws dep for the cloudwatch logs.

I had been hanging on to an updated ahash and crc32c dep locally due to a breakage with the stdsimd feature on linux. I didn't want to be too much a bother and push those, but the lack of updating that Cargo.lock caught up to me.

Also ran the dd-rust-license-tool to check licenses.

[jszwedko]

No worries! I kicked off CI again. Thanks for the additional changes!

[PriceHiller]

I'm unsure if those latest errors in the CI are things I can resolve.

Appears to be a result of a missing API key.