API

Use preheat apis for preheating. First create a POST request for preheating.

Configuration

If the client is containerd, it is recommended to configure proxy.defaultFilter in dfget.yaml and set it to Expires&Signature&ns, because containerd will add ns query params to the blobs download URL, refer to containerd/remotes/docker/resolver.go. Which will cause the generated Task ID to be different from the preheat Task ID, so it is impossible to hit the preheat blobs.

# proxy service detail option
proxy:
  defaultFilter: 'Expires&Signature&ns'

Operation

If the scheduler_cluster_ids does not exist, it means to preheat all scheduler clusters.

curl --location --request POST 'http://dragonfly-manager:8080/api/v1/jobs' \
--header 'Content-Type: application/json' \
--data-raw '{
    "type": "preheat",
    "args": {
        "type": "image",
        "url": "https://index.docker.io/v2/library/redis/manifests/latest"
    }
}'

The command-line log returns the preheat job id.

{
    "id": 1,
    "task_id": "group_4d1ea00e-740f-4dbf-a47e-dbdc08eb33e1",
    "type": "preheat",
    "status": "PENDING",
    "args": {
        "filter": "",
        "headers": null,
        "type": "image",
        "url": "https://index.docker.io/v2/library/redis/manifests/latest"
    }
}

Polling the preheating status with job id.

curl --request GET 'http://dragonfly-manager:8080/api/v1/jobs/1'

If the status is SUCCESS, the preheating is successful.

{
    "id": 1,
    "task_id": "group_4d1ea00e-740f-4dbf-a47e-dbdc08eb33e1",
    "type": "preheat",
    "status": "SUCCESS",
    "args": {
        "filter": "",
        "headers": null,
        "type": "image",
        "url": "https://index.docker.io/v2/library/redis/manifests/latest"
    }
}