• Back up PostgreSQL database and Redis data.
• Confirm Kubernetes cluster has sufficient resources for rolling update.

Example Helm values for Azure managed identity:

global:
  podLabels:
    azure.workload.identity/use: "true"

plugin_daemon:
  # azure managed service account
  serviceAccountName: "plugin-daemon-mi"

enterpriseAudit:
  serviceAccountName: "enterprise-audit-mi"

azureBlob:
  useManagedIdentity: true

Example per-database credential environment variables:

# If not set, will fallback to externalDatabase.username, externalDatabase.password
databaseCredentials:
  dify:
      user: ""
      password: ""
  enterprise:
      user: ""
      password: ""
  audit:
      user: ""
      password: ""
  plugin_daemon:
      user: ""
      password: ""

Pre-Upgrade Checklist

• Back up PostgreSQL database and Redis data
• Confirm Kubernetes cluster has sufficient resources for rolling update

Configuring Additional Workers (Optional)

The default Celery worker is not affected by this feature. The trigger queue has been merged into the common worker, so upgrading does not require a separate trigger worker and existing trigger functionality continues to work without any configuration changes.

You may optionally deploy additional workers that consume specific queues, giving you fine-grained control over resource allocation and improved processing throughput.

Add an additionalWorkers list to your Helm values.yaml:

additionalWorkers:
  - name: workflow-worker
    enabled: false
    replicas: 1
    celeryQueues: "workflow,workflow_storage,workflow_based_app_execution"
    celeryWorkerAmount: 2
    resources: {}
    nodeSelector: {}
    affinity: {}
    tolerations: []
    extraEnv: []