Nested Structures - YAML

Deep Nesting

While a simple name and value are easy to read, production-grade YAML for Kubernetes or Ansible can become incredibly deep.

Understanding the "Level" of your code is vital. Every time you indent further, you are creating a "Child" attribute for the "Parent" key above it.

1. Mappings of Mappings

This is used to describe objects that have sub-properties.

database:
  connection:
    host: 10.0.0.5
    port: 5432
  auth:
    username: admin
    secret_key: vault/db-key

In this structure:

database is the parent.
connection and auth are siblings.
host and port are the children of connection.

2. Lists of Mappings (Most Common)

In DevOps, we almost always deal with "Lists of Objects." For example, a Kubernetes Pod might run 3 different containers. Each container is an object (mapping) with a name and an image.

containers:                # A List
  - name: frontend         # Item 1 starts here (A Mapping)
    image: nginx:1.19
    ports:
      - containerPort: 80
  - name: logger           # Item 2 starts here (A Mapping)
    image: fluentd:latest

Common Mistake: Forgetting to indent the properties of a list item. The name and image keys belong to the list item started by the hyphen (-). If you don't line them up properly, the YAML won't parse.

3. Mappings of Lists

Sometimes a single object has multiple values for one attribute.

monitoring:
  targets:           # A scalar key pointing to a List
    - prometheus
    - jaeger
    - grafana
  alert_channels:    # Another scalar key pointing to a List
    - slack
    - email
    - pagerduty

The "Visual Test"

A great way to test your nesting is to draw vertical lines down the left side of your code.

Every line that starts at the same vertical column is a Sibling.
Every line that is further to the right is a Child.
Every line further to the left is a Parent (or belong to a different branch entirely).