P2P preheating integrates key P2P distribution capabilities of CNCF projects like Dragonfly (v1.0.5+) and Uber Kraken (v0.1.3+) into Harbor and allow users to define policies around this action.

Before preheating images from Harbor, you must first install a P2P engine in your environment. Refer to your P2P distribution engine’s installation guide for specific configuration steps.

Due to the limitations of the Kraken preheat API, there are extra configurations steps needed. Follow the Kraken configuration guide for more information on integrating Kraken and Harbor.

The system admin can create P2P preheat provider instances by providing preheat API endpoint of the selected vendor (Dragonfly or Kraken) and related credential if necessary. The created preheat provider instances can be used across all the projects.

The project admin can create multiple preheat policies under the specified project by setting the resource filters and preheat criteria (including: content trust and vulnerability situation) and choosing the P2P preheat provider instance added by the system administrator. The preheating policy can be triggered to start by manual, on a scheduled basis, or event-based ways. When the preheating policy is executing, all the images that match the criteria defined in the policy will be distributed to and cached in the target P2P engine for future pulling requests.

Harbor records each time a preheating policy is executed. You can check the details of preheating executions and the related logs from the Project’s page.

Please note that due to some historical reasons, there are two versions of Dragonfly, v1 and v2, and v1 has been archived and is no longer maintained, and v2 has a complete refactoring of v1, so v2 is not compatible with v1, the following is the version-compatible relationship between Harbor and Dragonfly, and it is recommended that you upgrade to the latest version of Dragonfly.

Pages in this section

• Manage Preheat Provider Instances