Customizing Labels and Annotations for Pods¶

Customizing Pod Labels¶

The Helm Chart allows you to customize labels for your Airflow objects. You can set global labels that apply to all objects and pods defined in the chart, as well as component-specific labels for individual Airflow components.

Global Labels¶

Global labels can be set using the labels parameter in your values file. These labels will be applied to all Airflow objects and pods defined in the chart:

values.yaml¶

labels:
  environment: production

Component-Specific Labels¶

You can also set specific labels for individual Airflow components, which will be merged with the global labels. Component-specific labels take precedence over global labels, allowing you to override them as needed.

For example, to add specific labels to different components:

values.yaml¶

# Global labels applied to all pods
labels:
  environment: production

# Scheduler specific labels
scheduler:
  labels:
    role: scheduler

# Worker specific labels
workers:
  labels:
    role: worker

# API Server specific labels
apiServer:
  labels:
    role: ui

Customizing Pod Annotations¶

Pod annotations can be customized similarly to labels using podAnnotations and airflowPodAnnotations.

Global Pod Annotations¶

Global pod annotations can be set using airflowPodAnnotations. These are applied to all Airflow component pods (scheduler, api-server/webserver, triggerer, dag-processor and workers):

values.yaml¶

airflowPodAnnotations:
  example.com/team: data-platform

Component-Specific Pod Annotations¶

Each component also supports its own podAnnotations. Component-specific annotations take precedence over global ones:

values.yaml¶

scheduler:
  podAnnotations:
    example.com/component: scheduler

Templated Pod Annotations¶

Both airflowPodAnnotations and podAnnotations support Helm template expressions. This allows annotations to reference release metadata or compute checksums of chart-managed resources, so that pods automatically restart when those resources change.

For example, to restart scheduler pods whenever the chart’s extra ConfigMaps change:

values.yaml¶

extraConfigMaps:
  my-listener-config:
    data: |
      listener.py: ...

scheduler:
  podAnnotations:
    checksum/extra-configmaps: '{{ include (print $.Template.BasePath "/configmaps/extra-configmaps.yaml") . | sha256sum }}'

You can also reference release metadata:

values.yaml¶

airflowPodAnnotations:
  release: '{{ .Release.Name }}'

Note

The include/sha256sum pattern only works for resources managed by this chart (e.g., those created via extraConfigMaps or extraSecrets). For ConfigMaps or Secrets created outside the chart, consider using a tool like Stakater Reloader to trigger pod restarts automatically.