»Variable Interpolation

Nomad supports interpreting two classes of variables: node attributes and runtime environment variables. Node attributes are interpretable in constraints, task environment variables, and certain driver fields. Runtime environment variables are not interpretable in constraints because they are only defined once the scheduler has placed them on a particular node.

The syntax for interpreting variables is ${variable}. An example and a comprehensive list of interpretable fields can be seen below:

task "docs" {
  driver = "docker"

  # Drivers support interpreting node attributes and runtime environment
  # variables
  config {
    image = "my-app"

    # Interpret runtime variables to inject the address to bind to and the
    # location to write logs to.
    args = [
      "--bind", "${NOMAD_ADDR_RPC}",
      "--logs", "${NOMAD_ALLOC_DIR}/logs",
    ]

    port_map {
      RPC = 6379
    }
  }

  # Constraints only support node attributes as runtime environment variables
  # are only defined after the task is placed on a node.
  constraint {
    attribute = "${attr.kernel.name}"
    value     = "linux"
  }

  # Environment variables are interpreted and can contain both runtime and
  # node attributes. These environment variables are passed into the task.
  env {
    "DC"      = "Running on datacenter ${node.datacenter}"
    "VERSION" = "Version ${NOMAD_META_VERSION}"
  }

  # Meta keys are also interpretable.
  meta {
    VERSION = "v0.3"
  }
}

»Node Variables

Below is a full listing of node attributes that are interpretable. These attributes are interpreted by both constraints and within the task and driver.

Below is a table documenting common node properties:

The full list of node attributes can be obtained by running nomad node status -verbose [node].

Here are some examples of using node attributes and properties in a job file:

job "docs" {
  # This will constrain this job to only run on 64-bit clients.
  constraint {
    attribute = "${attr.cpu.arch}"
    value     = "amd64"
  }

  # This will restrict the job to only run on clients with 4 or more cores.
  # Note: you may also declare a resource requirement for CPU for a task.
  constraint {
    attribute = "${cpu.numcores}"
    operator  = ">="
    value     = "4"
  }

  # Only run this job on a memory-optimized AWS EC2 instance.
  constraint {
    attribute = "${attr.platform.aws.instance-type}"
    value     = "m4.xlarge"
  }
}

»Environment Variables

The following are runtime environment variables that describe the environment the task is running in. These are only defined once the task has been placed on a particular node and as such can not be used in constraints.

Environment variables should be enclosed in brackets ${...} for interpolation.

»Dots in Variables

Starting in Nomad 0.9, task configuration interpolation requires variables to be valid identifiers. While this does not affect default variables or common custom variables, it is possible to define a variable that is not a valid identifier:

env {
  "valid.name"     = "ok"
  "invalid...name" = "not a valid identifier"
}

The environment variable invalid...name cannot be interpolated using the standard "${invalid...name}" syntax. The dots will be interpreted as object notation so multiple consecutive dots are invalid.

To continue supporting all user environment variables Nomad 0.9 added a new env variable which allows accessing any environment variable through index syntax:

task "redis" {
  driver = "docker"
  config {
    image  = "redis:3.2"
    labels {
      label1 = "${env["invalid...name"]}"
      label2 = "${env["valid.name"]}"
    }
  }
}