Configure metrics

Astro Private Cloud (APC) provides multiple options for collecting and exporting Airflow metrics including StatsD, OpenTelemetry (OTEL), and Prometheus integration.

StatsD configuration (default)

StatsD resource limits are managed at the API level via componentsConfig and apply to all components — they can’t be configured independently per component.

1
resources:
2
  requests:
3
    cpu: "100m"
4
    memory: "384Mi"
5
  limits:
6
    cpu: "100m"
7
    memory: "384Mi"

Airflow configuration

1
[metrics]
2
statsd_on = True
3
statsd_host = localhost
4
statsd_port = 8125
5
statsd_prefix = airflow

Prometheus integration

1
prometheus:
2
  enabled: true
3
  retention: 15d
4
  persistence:
5
    enabled: true
6
    size: 100Gi

Grafana dashboards

Access Grafana at:

https://grafana.<platform-domain>

Pre-built dashboards include:

• Airflow Dag performance
• Task execution metrics
• Scheduler health
• Worker utilization

Alerting

1
alertmanager:
2
  enabled: true
3
  config:
4
    route:
5
      receiver: 'platform'
6
    receivers:
7
      - name: 'platform'
8
        webhook_configs:
9
          - url: 'http://houston:8871/v1/alerts'

Built-in alerts

• AirflowDeploymentUnhealthy
• AirflowSchedulerUnhealthy
• AirflowTasksPendingIncreasing

Key metrics

Best practices

• Set appropriate retention based on storage capacity.
• Use OTEL for multi-backend export.
• Configure alerts for critical health metrics.
• Monitor task queue depth for scaling needs.