Welcome to Customer-Managed Prefect¶
Customer-Managed Prefect brings the Prefect Cloud API into your own infrastructure — giving your team full workflow orchestration with no external connectivity required.
Deploy & Manage¶
Resources for administrators deploying and operating Customer-Managed Prefect.
- Infrastructure Requirements — Kubernetes, PostgreSQL, Redis sizing and version requirements
- Deployment Guide — Step-by-step installation via Helm
- Architecture Overview — Cloud architecture and service components
- Upgrade Guide — How to upgrade between releases
- Release Notes — Latest versions and patch history
- Monitoring — Health checks and observability
- FAQ — Common operational questions
Reference¶
- Authentication Methods — SSO configuration (Okta, Entra, ADFS)
- Database Flags — Recommended PostgreSQL configuration
- Secrets Configuration — Managing secrets in your deployment
- Feature Flags — Enabling and disabling features
- Security — Security posture and encryption details
Troubleshooting¶
- Troubleshooting Runbook — Diagnosing common issues and building a support bundle
- Rollback Guide — Rolling back to a previous release
Get Started as a Developer¶
This guide will help you set up your local environment and deploy your first workflow to your instance.
Prerequisites¶
Before you begin, ensure you have:
- An active account on your customer-managed Prefect instance
- Python 3.10 or newer installed
Step 1: Install Prefect¶
Install the Prefect package:
pip install prefect
Check the releases page to see which version of Prefect is supported with your instance. If you need to install a specific version:
pip install "prefect<3.7"
Replace 3.7 with the next minor version above what your instance supports.
Step 2: Create a Prefect Profile¶
Prefect profiles allow you to manage multiple configurations. Create a new profile for your instance:
prefect profile create profile-name
Replace profile-name with a descriptive name (for example, my-company or staging).
Step 3: Configure Your Instance URLs¶
Configure the profile with your instance's URLs:
prefect config set PREFECT_CLOUD_UI_URL=https://app.your-instance.example.com
prefect config set PREFECT_CLOUD_API_URL=https://api.your-instance.example.com
Your instance administrator can provide these URLs if you don't have them.
Step 4: Generate and Configure Your API Key¶
Navigate to your settings in the Prefect UI and generate an API key. Then authenticate:
prefect cloud login -k <API_KEY>
You'll see a confirmation message when authentication is successful.
Work Pools¶
Customer-Managed Prefect cannot use managed execution work pool options. Before deploying flows, ensure you have access to a work pool. If you don't have one configured, contact your instance administrator or see the Prefect docs on work pools.
Next: Create Your First Workflow¶
Follow the Prefect Quickstart guide to create and deploy your first workflow. Your flows will run on your customer-managed instance instead of Prefect Cloud, and you'll deploy to your organization's self-hosted work pool.
Common Tasks¶
- Deploy a workflow — See the deployment guide
- Schedule runs — Learn about scheduling workflows
- Monitor runs — View and manage workflows in the Prefect UI at your configured
PREFECT_CLOUD_UI_URL - Configure work pools — Set up self-hosted work pools for your infrastructure
Troubleshooting¶
- Verify your instance URLs are correct and accessible
- Ensure your API key is valid (regenerate it in your settings if needed)
- View your current configuration:
prefect config view - Confirm that a work pool is available and configured for your deployments
For more help, refer to the full Prefect documentation or contact your instance administrator.