aws_mq_broker incorrect order of endpoints due to the new feature (prometheus monitoring)

[avoidik]

Terraform and AWS Provider Version

$ terraform --version
Terraform v1.15.1
on darwin_arm64

Affected Resource(s) or Data Source(s)

• aws_mq_broker

Expected Behavior

Correct order of endpoints

Actual Behavior

After AWS introduced Prometheus metrics support for Amazon MQ RabbitMQ (release notes), the broker endpoints list returned by the API now has the Prometheus metrics URL as the first entry. This breaks the Terraform AWS provider because it relies on positional index to resolve endpoints — what was previously endpoints[0] (AMQP) is now displaced.

Relevant Error/Panic Output

N/A

Sample Terraform Configuration

N/A

Steps to Reproduce

1. Deploy MQ broker by following https://registry.terraform.io/providers/hashicorp/aws/latest/docs/resources/mq_broker.html

Debug Logging

N/A

GenAI / LLM Assisted Development

N/A

Important Facts and References

Before maintenance:

$ aws mq describe-broker --query BrokerInstances[].Endpoints --broker-id abc
[
    [
        "amqps://url.on.aws:5671"
    ]
]

After maintenance:

$ aws mq describe-broker --query BrokerInstances[].Endpoints --broker-id abc
[
    [
        "https://url.on.aws:16001",
        "amqps://url.on.aws:5671"
    ]
]

Relevant issue #13786

Would you like to implement a fix?

No

[github-actions]

Community Guidelines

This comment is added to every new Issue to provide quick reference to how the Terraform AWS Provider is maintained. Please review the information below, and thank you for contributing to the community that keeps the provider thriving! 🚀

Voting for Prioritization

• Please vote on this Issue by adding a 👍 reaction to the original post to help the community and maintainers prioritize it.
• Please see our prioritization guide for additional information on how the maintainers handle prioritization.
• Please do not leave +1 or other comments that do not add relevant new information or questions; they generate extra noise for others following the Issue and do not help prioritize the request.

Volunteering to Work on This Issue

• If you are interested in working on this issue, please leave a comment.
• If this would be your first contribution, please review the contribution guide.
• For new resources and data sources, use skaff to generate scaffolding with comments detailing common expectations.

[dosubot]

Hey @avoidik 👋 Thank you for taking the time to raise this!

This is a valid issue. The aws_mq_broker resource copies endpoints directly from the AWS API response without any sorting or transformation [1], and since the endpoints attribute uses TypeList [2], positional changes in the API response are reflected directly in Terraform state. When AWS added the Prometheus metrics endpoint (https://...:16001) as the first entry, it shifted the AMQP endpoint from index 0 to index 1, breaking any configuration referencing instances.0.endpoints.0.

The provider's own test suite also assumes fixed positions for RabbitMQ brokers — specifically that instances.0.endpoints.0 matches amqps://...:5671 [3] — so this change would likely cause acceptance test failures as well.

A potential fix could involve sorting endpoints by protocol scheme (or port) within the flattenBrokerInstances function to ensure a stable order regardless of what AWS returns. Alternatively, the provider could introduce protocol-specific computed attributes (similar to how console_url is already broken out separately) to avoid positional indexing entirely.

Written by @dosu, approved by justinretzolk

To reply, just mention @dosu.

---

Share context across your team and agents. Try Dosu.

[avoidik]

Would it be okay if we reordered the returned list of API endpoints based on the protocol prefix, rather than assuming the list arrives pre-sorted as expected?

[avoidik]

I added a PR #47777