I've got a map variable that identifies existing s3 buckets:

resource "aws_s3_bucket" "bucket" {
  for_each = var.s3_replication
  bucket   = each.value.source
  #other configuration
}

variable "s3_replication" {
  description = "Map of buckets to replicate"
  type        = map
  default = {
    logs = {
      source = "logs_bucket",
      destination = "central_logs_bucket"
    },
    security = {
      source = "cloudtrail_bucket",
      destination = "central_security_bucket"
    }
  }
}

Since these buckets already exist, I am trying to import them and then apply the a configuration to them to update the resources. Unfortunately, I am not able to figure out how to do a terraform import on these. I've tried:

terraform import aws_s3_bucket.bucket["logs"] logs_bucket
terraform import aws_s3_bucket.bucket[logs] logs_bucket
terraform import aws_s3_bucket.bucket[0] logs_bucket
terraform import aws_s3_bucket.bucket[0].source logs_bucket
terraform import aws_s3_bucket.bucket[0[source]] logs_bucket

All failing with a different error. Any idea on how to import existing resources listed on a map?

CodePudding user response：

The terraform import subcommand relies on strings in the map key within resource namespaces that are first class expression, and this causes issues with shell interpreters where the resource is not a first class expression because they are not the Terraform DSL. You can work around this by casting the entire resource name as a literal string:

terraform import 'aws_s3_bucket.bucket["logs"]' logs_bucket

and this will resolve your issue.

CodePudding user response：

Before doing an import I recommend doing a terraform plan. The output of the plan would look like this:

Terraform will perform the following actions:

  # aws_s3_bucket.bucket["logs"] will be created
    resource "aws_s3_bucket" "bucket" {
        acceleration_status         = (known after apply)
        acl                         = "private"
        arn                         = (known after apply)
        bucket                      = "logs_bucket"
        bucket_domain_name          = (known after apply)
        bucket_regional_domain_name = (known after apply)
        force_destroy               = false
        hosted_zone_id              = (known after apply)
        id                          = (known after apply)
        region                      = (known after apply)
        request_payer               = (known after apply)
        tags_all                    = (known after apply)
        website_domain              = (known after apply)
        website_endpoint            = (known after apply)

        versioning {
            enabled    = (known after apply)
            mfa_delete = (known after apply)
        }
    }

  # aws_s3_bucket.bucket["security"] will be created
    resource "aws_s3_bucket" "bucket" {
        acceleration_status         = (known after apply)
        acl                         = "private"
        arn                         = (known after apply)
        bucket                      = "cloudtrail_bucket"
        bucket_domain_name          = (known after apply)
        bucket_regional_domain_name = (known after apply)
        force_destroy               = false
        hosted_zone_id              = (known after apply)
        id                          = (known after apply)
        region                      = (known after apply)
        request_payer               = (known after apply)
        tags_all                    = (known after apply)
        website_domain              = (known after apply)
        website_endpoint            = (known after apply)

        versioning {
            enabled    = (known after apply)
            mfa_delete = (known after apply)
        }
    }

Plan: 2 to add, 0 to change, 0 to destroy.

Having this plan we can see that the resources which will be created can be references as aws_s3_bucket.bucket["logs"] and aws_s3_bucket.bucket["security"]. Instead of doing an apply we can import them as follows:

Bash:

terraform import 'aws_s3_bucket.bucket["security"]' cloudtrail-bucket
terraform import 'aws_s3_bucket.bucket["logs"]' logs-bucket

Windows CMD:

terraform import 'aws_s3_bucket.bucket[\"security\"]' cloudtrail-bucket
terraform import 'aws_s3_bucket.bucket[\"logs\"]' logs-bucket

CodePudding user response：

The syntax from the answers posted are correct. My code was just bugging out because it couldn't locate the varsfile. I needed to add -var-file={{ path to tfvars }}. So the final syntax looks like:

terraform import -var-file={{ path to tfvars}} 'aws_s3_bucket.bucket["logs"]' logs_bucket