feat!: Upgrade AWS provider and min required Terraform version to 6.0 and 1.5.7 respectively

[bryantbiggs]

Description

List of backwards incompatible changes

• Terraform v1.5.7 is now minimum supported version
• AWS provider v6.0.0 is now minimum supported version
• instance_iam_role_additional_policies, service_iam_role_additional_policies, spot_fleet_iam_role_additional_policies types are now map(string) instead of list(string)
• IAM assume role policy SIDs have been modified to match their use (previously all were ECSAssumeRole which is inaccurate)
• compute_environment_order is now a required argument for aws_batch_job_queue per the Batch API and replaces compute_environments

Additional changes

Added

• Support for region parameter to specify the AWS region for the resources created if different from the provider region.
• Support for compute_environment_order, job_state_time_limit_action, timeouts arguments for job queues
• All (currently) supported arguments for eks_properties argument have been added to the job definition resource
• Support for scheduling_priority and node_properties arguments for job definitions

Modified

• Variable definitions now contain detailed object types in place of the previously used any type.
• compute_environment_name argument has been changed to name per provider v6.x API; no-op for users
• compute_environment_name_prefix argument has been changed to name_prefix per provider v6.x API; no-op for users

Removed

• None

Motivation and Context

• Resolves Invalid for_each argument when passing a IAM policy resource #15
• Resolves feat: Allow specifying compute environments for queue #16
• Resolves Removal of attempt_duration_seconds makes the terraform plan incoherent #21
• Resolves AWS Batch job definition always recreating if timeout is not provided. #22
• Resolves having scheduling priority available with the module in job definitions #26
• Resolves fix: Job definition is always recreated when attempt_duration_seconds is null #28
• Resolves fix: Use maps for IAM additional policies used in for_each #29
• Resolves Update Syntax for compute_environments #30
• Resolves feat: Parameter compute_environments set to deprecation by aws provider #31
• Resolves Invalid for_each argument when passing a IAM policy resource #34
• Resolves feat: Add support for update_policy #35
• Resolves Add Support for Types of Properties #37
• Resolves feat: Add support for aws_batch_job_queue.job_state_time_limit_action #39
• Closes fix: Rename deprecated variables #40

Breaking Changes

• Yes, see docs/UPGRADE-3.0.md for more details

How Has This Been Tested?

• I have updated at least one of the examples/* to demonstrate and validate my change(s)
• I have tested and validated these changes using one or more of the provided examples/* projects
• I have executed pre-commit run -a on my pull request

[bryantbiggs]

we'll proceed once AWS provider v6 arrives - this was originally meant to arrive in April but seems to be taking a bit longer

[antonbabenko]

Looks pretty good, just a small comment/question about variable types. Such detailed variable types are a rather big change for us in the modules.

[antonbabenko]

Such a detailed variable type may be a good idea, but do we need to provide a default value ("MANAGED") in the type definition or leave it empty, so that it will be null and let the AWS provider decide on the correct value? We previously left it empty for the provider to decide. I think it is still right to do.

[bryantbiggs]

yes, the standard practices still apply and null should be favored in 99% of the cases. in this instance, "MANAGED" was specifically chosen and is also what we have on the current module version

terraform-aws-batch/main.tf

Line 14 in c478369

type = lookup(each.value, "type", "MANAGED")

For this argument users have to choose either "MANAGED" or "UNMANAGED", we've made the decision that is most common for them as the default

[antonbabenko]

Same here:

Suggested change

	terminate_jobs_on_update = optional(bool, false)
	terminate_jobs_on_update = optional(bool)

[bryantbiggs]

This is a new argument thats added in this PR that is required so we are making a "sane" default for users (i.e. - don't terminate the job if you update the compute environment, instead let the job complete - which is what most will want to do in order to minimize disruptions for changes)

[antonbabenko]

This PR is included in version 3.0.0 🎉

[github-actions]

I'm going to lock this pull request because it has been closed for 30 days ⏳. This helps our maintainers find and focus on the active issues. If you have found a problem that seems related to this change, please open a new issue and complete the issue template so we can capture all the details necessary to investigate further.