Kubernetes

Dragonfly Quick Start document aims to help you to quick start Dragonfly journey. This experiment is quite easy and simplified.

You can have a quick start following Helm Charts. We recommend to use Containerd with CRI and CRI-O client.

This table describes some container runtimes version and documents.

Runtime	Version	Document	CRI Support	Pull Command
Containerd*	v1.1.0+	Link	Yes	crictl pull docker.io/library/alpine:latest
Containerd without CRI	v1.1.0	Link	No	ctr image pull docker.io/library/alpine
CRI-O	All	Link	Yes	crictl pull docker.io/library/alpine:latest

containerd is recommended.

Runtime Configuration Guide for Dragonfly Helm Chart

Dragonfly helm supports config docker automatically.

Config cases:

Implicit registries support

Chart customize values.yaml:

containerRuntime:
  docker:
    enable: true
    # -- Inject domains into /etc/hosts to force redirect traffic to dfdaemon.
    # Caution: This feature need dfdaemon to implement SNI Proxy,
    # confirm image tag is greater than v0.4.0.
    # When use certs and inject hosts in docker, no necessary to restart docker daemon.
    injectHosts: true
    registryDomains:
      - 'harbor.example.com'
      - 'harbor.example.net'

This config enables docker pulling images from registries harbor.example.com and harbor.example.net via Dragonfly. When deploying Dragonfly with above config, it's unnecessary to restart docker daemon.

Limitations:

• Only support implicit registries

Prepare Kubernetes Cluster

If there is no available Kubernetes cluster for testing, minikube is recommended. Just run minikube start.

Install Dragonfly

Install with default configuration

helm repo add dragonfly https://dragonflyoss.github.io/helm-charts/
helm install --create-namespace --namespace dragonfly-system dragonfly dragonfly/dragonfly

Wait Dragonfly Ready

Wait all pods running

kubectl -n dragonfly-system wait --for=condition=ready --all --timeout=10m pod

Manager Console

The console page will be displayed on dragonfly-manager.dragonfly-system.svc.cluster.local:8080.

If you need to bind Ingress, you can refer to configuration options of Helm Charts, or create it manually.

Console features preview reference document console preview.

Using Dragonfly

After all above steps, create a new pod with target registry. Or just pull an image with crictl:

crictl harbor.example.com/library/alpine:latest

crictl pull docker.io/library/alpine:latest

After pulled images, find logs in dfdaemon pod:

# find pods
kubectl -n dragonfly-system get pod -l component=dfdaemon
# find logs
pod_name=dfdaemon-xxxxx
kubectl -n dragonfly-system exec -it ${pod_name} -- grep "peer task done" /var/log/dragonfly/daemon/core.log

Example output:

{
  "level": "info",
  "ts": "2022-09-07 12:04:26.485",
  "caller": "peer/peertask_conductor.go:1500",
  "msg": "peer task done, cost: 1ms",
  "peer": "10.140.2.175-5184-1eab18b6-bead-4b9f-b055-6c1120a30a33",
  "task": "b423e11ddb7ab19a3c2c4c98e5ab3b1699a597e974c737bb4004edeef6016ed2",
  "component": "PeerTask"
}

Preheat

To get the best out of Dragonfly, you can pull the image in advance by preheat