Target Plugins

Target Plugins determine where the resource to be autoscaled is located. All target plugins support the dry-run policy config parameter which allows a policy to be evaluated, but will noop any suggested changes.

Below is a list of plugins you can use with the Nomad Autoscaler:

Nomad Task Group Target

The Nomad task group target indicates the scalable resource is a Nomad job running on a Nomad cluster.

Agent Configuration Options

The Nomad target is automatically launched by the Nomad Autoscaler and so the following setup is optional.

target "nomad" {
  driver = "nomad"
}

Policy Configuration Options

If using the Nomad job specification scaling stanza to configure the scaling policy, the following section can be omitted as Nomad will populate them on job submission.

check "example-check" {
  ...
  target "nomad" {
    Namespace = "default"
    Job       = "example"
    Group     = "cache"
  }
  ...

Namespace (string: "") - The namespace in which the job resides as defined by the namespace parameter within the job specification.
Job (string: "") - The job identifier which contains the task group to scale as defined within the job specification job stanza.
Group (string: "") - The name of the task group to scale as defined in the job specification group stanza.

Dynamic Application Sizing Nomad Task Target

This functionality only exists in Nomad Autoscaler Enterprise. This is not present in the open source version of Nomad Autoscaler.

The app-sizing-nomad target plugin reports on the current task resource value as well as submits recommendations to Nomad via the recommendations API endpoint based on the result of the Dynamic Application Sizing strategy calculations.

Agent Configuration Options

The app-sizing-nomad plugin is automatically launched by Nomad Autoscaler Enterprise and so the following setup is optional.

target "app-sizing-nomad" {
  driver = "app-sizing-nomad"
}

Policy Configuration Options

If using the Nomad job specification scaling stanza to configure the scaling policy for a task resource, the following section can be omitted as Nomad will populate them on job submission. Please also refer to the Nomad Autoscaler policy documentation for additional detail and guidance.

check "mem" {
  ...
  target "app-sizing-nomad" {
    Namespace = "default"
    Job       = "example"
    Group     = "cache"
    Task      = "Redis"
    Resource  = "MemoryMB"
  }
  ...

Region (string: "") - The region in which the job resides as defined by the region parameter within the job specification.
Namespace (string: "") - The namespace in which the job resides as defined by the namespace parameter within the job specification.
Job (string: "") - The job identifier which contains the task group to scale as defined within the job specification job stanza.
Group (string: "") - The name of the task group to scale as defined in the job specification group stanza.
Task (string: "") - The name of the task to scale as defined in the job specification task stanza.
Resource (string: "") - The resource to scale as defined by the job specification resource stanza. Supports either CPU or MemoryMB.

AWS AutoScaling Group Target

The AWS ASG target plugin allows for the scaling of the Nomad cluster clients via manipulating AWS AutoScaling Groups.

Agent Configuration Options

To use the AWS ASG target plugin, the agent configuration needs to be populated with the appropriate target block. Authentication to the AWS API can be supplied in a number of ways including EC2 instance roles. It is recommended, if possible to use the Vault AWS Secrets engine for supplying access credentials to the plugin. Credentials should be injected into the configuration via a template rather than as environment variables. This ensures the credentials are passed only to the plugin, rather than being available for all plugins and the agent process.

The IAM policy required for the AWS ASG plugin to function properly is detailed below.

{
  "Version": "2012-10-17",
  "Statement": [
    {
      "Sid": "",
      "Effect": "Allow",
      "Action": [
        "ec2:TerminateInstances",
        "ec2:DescribeInstanceStatus",
        "autoscaling:UpdateAutoScalingGroup",
        "autoscaling:DetachInstances",
        "autoscaling:DescribeScalingActivities",
        "autoscaling:DescribeAutoScalingGroups",
        "autoscaling:CreateOrUpdateTags"
      ],
      "Resource": "*"
    }
  ]
}

target "aws-asg" {
  driver = "aws-asg"
  config = {
    aws_region            = "eu-west-3"
    aws_access_key_id     = "AKIAIOSFODNN7EXAMPLE"
    aws_secret_access_key = "wJalrXUtnFEMI/K7MDENG/bPxRfiCYEXAMPLEKEY"
  }
}

When using a Nomad cluster with ACLs enabled, the plugin will require an ACL token which provides the following permissions:

node {
  policy = "write"
}

aws_region (string: "us-east-1") - The AWS region identifier to connect to and where resources should be managed.
aws_access_key_id (string: "") - The AWS access key ID used to authenticate with the AWS API.
aws_secret_access_key (string: "") - The AWS secret key ID used to authenticate with the AWS API.
aws_session_token (string: "") - The AWS session token used to authenticate with the AWS API.

Policy Configuration Options

check "hashistack-allocated-cpu" {
  ...
  target "aws-asg" {
    aws_asg_name        = "hashistack-client-asg"
    node_class          = "hashistack"
    node_drain_deadline = "5m"
    node_purge          = "true"
  }
  ...

aws_asg_name (string: <required>) - The name of the AWS AutoScaling Group to interact with when performing scaling actions.
node_class (string: <required>) - The Nomad client node class identifier used to group nodes into a pool of resource.
node_drain_deadline (duration: "15m") The Nomad drain deadline to use when performing node draining actions.
node_purge (bool: "false") A boolean flag to determine whether Nomad clients should be purged when performing scale in actions.

Azure Virtual Machine Scale Set Target

The Azure VMSS target plugin allows for the scaling of the Nomad cluster clients via manipulating Azure Virtual Machine Scale Sets.

The Azure VMSS target plugin requires Azure fingerprinting, which became available in Nomad 1.0. If you are running a previous version of Nomad you will to set a meta tag in all of your nodes.

Agent Configuration Options

To use the Azure VMSS target plugin, the agent configuration needs to be populated with the appropriate target block. Authentication to the Azure API can be supplied in a number of ways.

Virtual Machine Identities

When using virtual machine identities you will need to provide a user-managed identity with the Contributor role and set the subscription_id in the Autoscaler configuration file:

target "azure-vmss" {
  driver = "azure-vmss"
  config = {
    subscription_id   = "ee0886ab-5cc2-4583-a3f0-c4bfd044ee82"
  }
}

Configuration file

Credentials should be injected into the configuration via a template rather than as environment variables. This ensures the credentials are passed only to the plugin, rather than being available for all plugins and the agent process. It is recommended, if possible to use the Vault Azure Secrets engine for supplying access credentials to the plugin.

target "azure-vmss" {
  driver = "azure-vmss"
  config = {
    tenant_id         = "02e99e69-91b0-4d0a-bd61-bf5c08fbfbab"
    client_id         = "34c276fa-2a74-4cf9-9fce-4e60642274cb"
    secret_access_key = "ThisIrznTIS_~FAKEj4X93FTyAgilUC511"
    subscription_id   = "ee0886ab-5cc2-4583-a3f0-c4bfd044ee82"
  }
}

tenant_id (string: "") - The ID of the tenant to authenticate to.
client_id (string: "") - The app ID of the user-assigned identity in Azure Active Directory.
secret_access_key (string: "") - The secret key ID used to authenticate with the Azure API.
subscription_id (string: "") - The ID of the subscription to authenticate to.

Nomad ACL

When using a Nomad cluster with ACLs enabled, the plugin will require an ACL token which provides the following permissions:

node {
  policy = "write"
}

Policy Configuration Options

check "clients-azure-vmss" {
  ...
  target "azure-vmss" {
    resource_group      = "prod"
    vm_scale_set        = "hashistack-client-set"
    node_class          = "hashistack"
    node_drain_deadline = "5m"
    node_purge          = "true"
  }
  ...

resource_group (string: <required>) - The name of the Azure resource group within which the virtual machine scale set resides.
vm_scale_set (string: <required>) - The name of the Azure virtual machine scale set to interact with when performing scaling actions.
node_class (string: <required>) - The Nomad client node class identifier used to group nodes into a pool of resource.
node_drain_deadline (duration: "15m") The Nomad drain deadline to use when performing node draining actions.
node_purge (bool: "false") A boolean flag to determine whether Nomad clients should be purged when performing scale in actions.

Client meta tag

The Azure VMSS target plugin requires Nomad 1.0 for Azure fingerprinting. Alternatively, you can set a client meta tag called unique.platform.azure.name with the value being the Azure VMSS node name. This can be retrieved programatically via the Azure Instance Metadata service.