lancecwhite commented 1 week ago

Terraform Core Version

1.8.5

AWS Provider Version

5.65.0

Affected Resource(s)

When increasing the node count value cluster_cache_nodes for aws_elasticache_cluster the output appears to be delayed to only updating during the next execution of terraform apply. So it appears that the resource isn't making a request against the managed service to output the updated list of nodes.

Expected Behavior

expected output for cluster_cache_nodes... cluster_cache_nodes = tolist([ { "address" = "test-memcache.xxxxx.0001.use1.cache.amazonaws.com" "availability_zone" = "us-east-1a" "id" = "0001" "outpost_arn" = "" "port" = 11211 }, { "address" = "test-memcache.xxxxx.0002.use1.cache.amazonaws.com" "availability_zone" = "us-east-1b" "id" = "0002" "outpost_arn" = "" "port" = 11211 }, { "address" = "test-memcache.xxxxx.0003.use1.cache.amazonaws.com" "availability_zone" = "us-east-1c" "id" = "0003" "outpost_arn" = "" "port" = 11211 }, { "address" = "test-memcache.xxxxx.0004.use1.cache.amazonaws.com" "availability_zone" = "us-east-1a" "id" = "0004" "outpost_arn" = "" "port" = 11211 }, ])

Actual Behavior

As an example, if you modify the node count for cluster_cache_nodes from 2 to 4, the output after the apply finishes remains: cluster_cache_nodes = tolist([ { "address" = "test-memcache.xxxxx.0001.use1.cache.amazonaws.com" "availability_zone" = "us-east-1a" "id" = "0001" "outpost_arn" = "" "port" = 11211 }, { "address" = "test-memcache.xxxxx.0002.use1.cache.amazonaws.com" "availability_zone" = "us-east-1b" "id" = "0002" "outpost_arn" = "" "port" = 11211 }, ])

Relevant Error/Panic Output Snippet

No response

Terraform Configuration Files

https://registry.terraform.io/providers/hashicorp/aws/latest/docs/resources/elasticache_cluster

Steps to Reproduce

terraform apply

Debug Output

No response

Panic Output

No response

Important Factoids

No response

References

https://registry.terraform.io/providers/hashicorp/aws/latest/docs/resources/elasticache_cluster

Would you like to implement a fix?

None

github-actions[bot] commented 1 week ago

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 commented 1 week ago

Hey @lancecwhite 👋 Thank you for taking the time to raise this! Are you able to supply a sample configuration that can be used to reproduce this, and/or debug logs (redacted as needed)? That sort of information is necessary for whoever ultimately picks this up to work on it.

One immediate question I have is what value you're setting for the apply_immediately argument. There's a note near the top of the resource documentation that mentions:

When you change an attribute, such as num_cache_nodes, by default it is applied in the next maintenance window. Because of this, Terraform may report a difference in its planning phase because the actual modification has not yet taken place. You can use the apply_immediately flag to instruct the service to apply the change immediately.

lancecwhite commented 1 week ago

Hey @lancecwhite 👋 Thank you for taking the time to raise this! Are you able to supply a sample configuration that can be used to reproduce this, and/or debug logs (redacted as needed)? That sort of information is necessary for whoever ultimately picks this up to work on it.

One immediate question I have is what value you're setting for the apply_immediately argument. There's a note near the top of the resource documentation that mentions:

When you change an attribute, such as num_cache_nodes, by default it is applied in the next maintenance window. Because of this, Terraform may report a difference in its planning phase because the actual modification has not yet taken place. You can use the apply_immediately flag to instruct the service to apply the change immediately.

Hey @justinretzolk, thanks for the reply. So currently, I am default apply_immediately to "true" for the module.

Here is a sample of our module execution:

module "memcached" {
  source                      = "./modules/aws/elasticache"
  apply_immediately           = var.apply_immediately
  az_mode                     = var.az_mode
  cluster_id                  = var.cluster_id
  create                      = var.create
  create_cluster              = var.create_cluster
  create_parameter_group      = var.create_parameter_group
  create_replication_group    = false # ensures memcached deployment rather than redis
  create_subnet_group         = var.create_subnet_group
  description                 = var.description
  engine                      = "memcached"
  engine_version              = var.engine_version
  maintenance_window          = var.maintenance_window
  network_type                = var.network_type
  node_type                   = var.node_type
  num_cache_nodes             = var.num_cache_nodes
  parameter_group_description = var.parameter_group_description
  parameter_group_family      = var.parameter_group_family
  parameter_group_name        = var.parameter_group_name
  parameters                  = var.memcached_parameters
  port                        = var.port
  security_group_ids          = var.security_group_ids
  subnet_group_description    = var.subnet_group_description
  subnet_group_name           = var.subnet_group_name
  subnet_ids                  = var.subnet_ids
  tags                        = var.tags
  transit_encryption_enabled  = var.transit_encryption_enabled
  vpc_id                      = var.vpc_id
}

justinretzolk commented 1 week ago

Hey @lancecwhite,

Are you able to supply the configuration of the aws_elasticache_cluster resource within the module and where var. apply_immediately is being set? It might also be helpful to know how the cluster_cache_nodes output is defined in the configuration (I'm assuming it's just referencing the related attribute of aws_elasticache_cluster, but being explicit is helpful for whoever winds up picking this up).

lancecwhite commented 1 week ago

Hey @lancecwhite,

Are you able to supply the configuration of the aws_elasticache_cluster resource within the module and where var. apply_immediately is being set? It might also be helpful to know how the cluster_cache_nodes output is defined in the configuration (I'm assuming it's just referencing the related attribute of aws_elasticache_cluster, but being explicit is helpful for whoever winds up picking this up).

@justinretzolk

Here is an output of my local.auto.tfvars used during the terraform apply:

additional_ingress_rules = [{"description":"Ingress - xxxx - Memcached","from_port":11211,"to_port":11211,"ip_protocol":"tcp","cidr_ipv4":"x.x.x.x/20"}]
apply_immediately = true
aws_profile = "null-memcached-manager"
aws_region = "us-east-1"
cluster_id = "null-test"
default_tags = {"JobName":"Memcached-null-test","SystemName":"null","TFModule":"Memcached","TFStatePath":"s3://null-temp-s3-bucket-tf-backend/null-Memcached-null.tfstate"}
description = "null-test cluster"
engine_version = "1.6.17"
maintenance_window = "mon:09:30-mon:10:30"
memcached_parameters = []
node_type = "cache.r7g.large"
num_cache_nodes = 2
parameter_group_family = "memcached1.6"
parameter_group_name = "null-test-parameter-group"
subnet_group_description = "Subnet group for Memcached - null-test"
subnet_group_name = "null-test-subnet-group"
subnet_ids = ["subnet-xxxxxx","subnet-xxxxxxx"]
tags = {"ud:rsc:product-name":"uem memcached","ud:rsc:environment":"dev","ud:uem:datacenter":"null"}
transit_encryption_enabled = false
vpc_id = "vpc-xxxxxxx"

Could you provide a bit more detail on what's needed for cluster_cache_nodes? The output seems to just trail the actual deployment of the resource. So in the example I provided the num_cache_nodes is being set to 4 from a value of 2. I would expect to see the below:

      cluster_cache_nodes:
        - address: "test-memcache.xxxxx.0001.use1.cache.amazonaws.com"
          availability_zone: "us-east-1a"
          id: "0001"
          outpost_arn: ""
          port: 11211
        - address: "test-memcache.xxxxx.0002.use1.cache.amazonaws.com"
          availability_zone: "us-east-1b"
          id: "0002"
          outpost_arn: ""
          port: 11211
        - address: "test-memcache.xxxxx.0003.use1.cache.amazonaws.com"
          availability_zone: "us-east-1c"
          id: "0003"
          outpost_arn: ""
          port: 11211
        - address: "test-memcache.xxxxx.0004.use1.cache.amazonaws.com"
          availability_zone: "us-east-1a"
          id: "0004"
          outpost_arn: ""
          port: 11211

But even with the apply_immediately this is what I see for the output:

      cluster_cache_nodes:
        - address: "test-memcache.xxxxx.0001.use1.cache.amazonaws.com"
          availability_zone: "us-east-1a"
          id: "0001"
          outpost_arn: ""
          port: 11211
        - address: "test-memcache.xxxxx.0002.use1.cache.amazonaws.com"
          availability_zone: "us-east-1b"
          id: "0002"
          outpost_arn: ""
          port: 11211

Until the NEXT attempt to run a modification against the cluster where changes to the output now includes an addition of the two resources that were added during the previous apply.

hashicorp / terraform-provider-aws

[Bug]: delayed cluster_cache_nodes output on aws_elasticache_cluster resource #39122

Terraform Core Version

AWS Provider Version

Affected Resource(s)

Expected Behavior

Actual Behavior

Relevant Error/Panic Output Snippet

Terraform Configuration Files

Steps to Reproduce

Debug Output

Panic Output

Important Factoids

References

Would you like to implement a fix?

Community Note