[Bug]: error creating ElastiCache Replication Group: InvalidParameterCombination: Expected a parameter group of family redis7 but found one of family redis6.x

[wgaldino]

Terraform Core Version

1.3.4

AWS Provider Version

4.38.0, 4.37.0, 4.36.0, 4.35.0, 4.34.0

Affected Resource(s)

• aws_elasticache_replication_group

Expected Behavior

I noticed that since yesterday any aws_elasticache_replication_group resource that before could have been created with the code below was failing.

resource "aws_elasticache_replication_group" "service" {
  count = local.count

  replication_group_id = "service-${local.some_sufix}"
  description          = "Service Regional Redis cluster"
  engine               = "redis"
  node_type = var.node_type
  automatic_failover_enabled = true

  parameter_group_name = "default.redis6.x.cluster.on"

  subnet_group_name  = aws_elasticache_subnet_group.service[0].name
  security_group_ids = [data.aws_security_group.default.id]
  num_node_groups    = var.node_count
  apply_immediately  = true
}

Actual Behavior

Apply fails with the following message:

Error: error creating ElastiCache Replication Group: InvalidParameterCombination: Expected a parameter group of family redis7 but found one of family redis6.x

Relevant Error/Panic Output Snippet

No response

Terraform Configuration Files

resource "aws_elasticache_replication_group" "service" {
  count = local.count

  replication_group_id = "service-${local.some_sufix}"
  description          = "Service Regional Redis cluster"
  engine               = "redis"
  node_type = var.node_type
  automatic_failover_enabled = true

  parameter_group_name = "default.redis6.x.cluster.on"

  subnet_group_name  = aws_elasticache_subnet_group.service[0].name
  security_group_ids = [data.aws_security_group.default.id]
  num_node_groups    = var.node_count
  apply_immediately  = true
}

Steps to Reproduce

Terraform init Terraform apply

Debug Output

No response

Panic Output

No response

Important Factoids

Today after some investigation I could extract the following facts:

• In 1 (eu-central-1) out of 4 regions succeeded and all resources got successfully created.
• Doesn't matter which version of the provider is used the outcome remains the same.
• The documentation is somehow outdated for this edge case since it implies that engine_version is optional, however if you are using the parameter_group_name argument it will fail. Mainly because if the engine_version is not defined it will get the latest (redis7) and that will clash with whatever is set in parameter_group_name.

The way to solve this issue was to add the engine_version argument to the resource definition. That's exactly the behaviour you get when creating the cluster through the GUI, you cannot set parameter_group_name as default.redis6.x.cluster.on if you select a different engine_version.

Possible ways of fixing it would be:

1. Update the documentation so that is states that you cannot arbitrarily set parameter_group_name when engine_version is omitted. Enforce that those need to match somehow.
2. Since the documentation states that engine_version is optional, why not infer the engine based on the parameter_group_name value? Doesn't it make sense? Is that really tricky?

References

No response

Would you like to implement a fix?

No response

[github-actions[bot]]

Community Note

Voting for Prioritization

• Please vote on this issue by adding a 👍 reaction to the original post to help the community and maintainers prioritize this request.
• Please see our prioritization guide for information on how we prioritize.
• Please do not leave "+1" or other comments that do not add relevant new information or questions, they generate extra noise for issue followers 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.

[justinretzolk]

Related #27709

[github-actions[bot]]

Marking this issue as stale due to inactivity. This helps our maintainers find and focus on the active issues. If this issue receives no comments in the next 30 days it will automatically be closed. Maintainers can also remove the stale label.

If this issue was automatically closed and you feel this issue should be reopened, we encourage creating a new issue linking back to this one for added context. Thank you!