ec_deployment: Support Autoscaling

[marclop]

Overview

Add support for the new autoscaling feature. API Ref.

Possible Implementation

resource "ec_deployment" "autoscale" {
  name = "autoscaling_deployment"

  # Mandatory fields
  region                 = "us-east-1"
  version                = "7.11.2"
  deployment_template_id = "aws-io-optimized-v2"

  elasticsearch {
      topology {
        id = "hot_content"
        # Optional: Default value "memory"
        size_resource = "memory" 
        
        # Current for the deployment, if autoscaling kicks in, this will
        # change in the deployment status, meaning that if "terraform plan"
        # runs and the deployment has autoscaled, there will be a diff on this
        # field.
        size = "32g"

        # Optional
        autoscale_max {
            # Optional: Default value "memory"
            size_resource = "memory" 
            
            # Required
            size = "128g"
        }
      }

      topology {
        id = "warm"
        
        # Optional: Default value "memory"
        size_resource = "memory" 
        
        # Current for the deployment, if autoscaling kicks in, this will
        # change in the deployment status, meaning that if "terraform plan"
        # runs and the deployment has autoscaled, there will be a diff on this
        # field.
        size = "64g"

        # Optional
        autoscale_max {
            # Optional: Default value "memory"
            size_resource = "memory" 
            
            # Required
            size = "128g"
        }
      }

      topology {
        id = "cold"

        # Optional: Default value "memory"
        size_resource = "memory" 
        
        # Current for the deployment, if autoscaling kicks in, this will
        # change in the deployment status, meaning that if "terraform plan"
        # runs and the deployment has autoscaled, there will be a diff on this
        # field.
        size = "128g"

        autoscale_max {
            # Optional: Default value "memory"
            resource = "memory" 
            
            # Required
            size = "256g"
        }
      }

      topology {
        id = "ml"
        
        # Optional
        autoscale_max {
            # Optional: Default value "memory"
            resource = "memory" 
            
            # Required
            size = "128g"
        }

        # For Phase 1 only ML supports autoscale_min. This is validated by the API.
        autoscale_min {
            # Optional: Default value "memory"
            resource = "memory"

            # Required
            size = "0g"
        }
      }
  }

  kibana {}
}

[Kushmaro]

@marclop how about having a single "autoscaling" object, with min/max values, rather two separate objects per topology?

[marclop]

how's this look?

resource "ec_deployment" "autoscale" {
  name = "autoscaling_deployment"

  # Mandatory fields
  region                 = "us-east-1"
  version                = "7.11.2"
  deployment_template_id = "aws-io-optimized-v2"

  elasticsearch {
      topology {
        id = "hot_content"
        # Optional: Default value "memory"
        size_resource = "memory" 
        
        # Current for the deployment, if autoscaling kicks in, this will
        # change in the deployment status, meaning that if "terraform plan"
        # runs and the deployment has autoscaled, there will be a diff on this
        # field.
        size = "32g"

        # Optional
        autoscale {
            # Optional: Default value "memory"
            max_size_resource = "memory" 
            
            # Required
            max_size = "128g"
        }
      }

      topology {
        id = "warm"
        
        # Optional: Default value "memory"
        size_resource = "memory" 
        
        # Current for the deployment, if autoscaling kicks in, this will
        # change in the deployment status, meaning that if "terraform plan"
        # runs and the deployment has autoscaled, there will be a diff on this
        # field.
        size = "64g"

        # Optional
        autoscale {
            # Optional: Default value "memory"
            max_size_resource = "memory" 
            
            # Required
            max_size = "128g"
        }
      }

      topology {
        id = "cold"

        # Optional: Default value "memory"
        size_resource = "memory" 
        
        # Current for the deployment, if autoscaling kicks in, this will
        # change in the deployment status, meaning that if "terraform plan"
        # runs and the deployment has autoscaled, there will be a diff on this
        # field.
        size = "128g"

        autoscale {
            # Optional: Default value "memory"
            max_size_resource = "memory" 
            
            # Required
            max_size = "256g"
        }
      }

      topology {
        id = "ml"
        
        # Optional
        autoscale {
            # Optional: Default value "memory"
            max_resource = "memory" 
            # Required
            max_size = "128g"

            # For Phase 1 only ML supports autoscale_min. This is validated by the API.
            # Optional: Default value "memory"
            min_resource = "memory"

            # Required
            min_size = "0g"
        }
      }
  }

  kibana {}
}

[karencfv]

Hmmmm... I'm not too sold on having the top level "global" autoscaling object.

It's a fact that we have to have per topology autoscaling objects. Having an additional "global" object will only add complexity to the schema in my opinion. The API also doesn't have this ability so we should tread carefully I think 😄

Please ignore this comment, I saw something that wasn't there. (Friday brain 😅 )