您可以在一个 K8S 集群中快速尝试 Fedlearner 的基础功能,部署 Fedlearner 会在您的 K8S 集群中安装以下组件:
- fedlearner-stack 包含了 Fedlearner 所依赖的基础设施,包括 NFS Server/RDS (mariadb) 等;
- fedleaner 包含了运行 Fedlearner 任务所需的组件,包括 Fedlearner Operator/Fedlearner APIServer 和 Fedlearner WebConsole
请注意,Fedlearner Operator 目前仅适用于 K8S 1.14.8 - 1.16.9 版本 (推荐 1.16.9 的 K8S 版本),我们正在适配更新版本的 K8S。
Kubernetes 集群中的节点必须安装 nfs-common
依赖,例如:
apt install nfs-common
helm install fedlearner-stack ./deploy/charts/fedlearner-stack
cat <<EOF | kubectl apply -f -
apiVersion: v1
kind: PersistentVolumeClaim
metadata:
name: pvc-fedlearner-default
spec:
accessModes:
- ReadWriteMany
resources:
requests:
storage: 10Gi
storageClassName: nfs
EOF
export OPERATOR_IMAGE=`git rev-parse HEAD | cut -c 1-7`
helm install fedlearner ./deploy/charts/fedlearner \
--set fedlearner-operator.ingress.enabled=false \
--set fedlearner-operator.image.tag=$OPERATOR_IMAGE \
--set fedlearner-operator.extraArgs.ingress-enabled-client-auth=false
kubectl create ns follower
helm install fedlearner ./deploy/charts/fedlearner --namespace follower \
--set fedlearner-apiserver.enabled=false \
--set fedlearner-web-console.enabled=false \
--set fedlearner-operator.installCRD=false \
--set fedlearner-operator.image.tag=$OPERATOR_IMAGE \
--set fedlearner-operator.extraArgs.ingress-enabled-client-auth=false
cat <<EOF | kubectl apply -f -
apiVersion: v1
kind: Secret
metadata:
name: test-secret
namespace: default
stringData:
ca.crt: fake
---
apiVersion: v1
kind: Secret
metadata:
name: test-secret
namespace: follower
stringData:
ca.crt: fake
EOF
kubectl create -f ./deploy/examples/normal_leader.yaml -f ./deploy/examples/normal_follower.yaml