1. 下载tf-operator所需的docker镜像，可能需要科学上网，每个k8s节点都需要pull该docker镜像

docker pull gcr.io/kubeflow-images-public/tf_operator:vmaster-ga2ae7bff

不能科学上网的同学，点击此处免费下载tf-operator.tar
下载完成后，使用docker load，将镜像load到集群的各个节点

docker load <  tf-operator.tar

2. 在kubeflow官网找到下载kfctl的地址，下载 kfctl和 kfctl_k8s_istio.v1.1.0.yaml

wget https://github.com/kubeflow/kfctl/releases/download/v1.1.0/kfctl_v1.1.0-0-g9a3621e_linux.tar.gz
tar -zxvf kfctl_v1.1.0-0-g9a3621e_linux.tar.gz

下载kfctl_k8s_istio.v1.1.0.yaml

wget https://raw.githubusercontent.com/kubeflow/manifests/v1.1-branch/kfdef/kfctl_k8s_istio.v1.1.0.yaml
kfctl build -V -f kfctl_k8s_istio.v1.1.0.yaml

cd  .cache/manifests/manifests-1.1-branch/tf-training

3. 下载kustomize 并执行kustomize build

wget https://github.com/kubernetes-sigs/kustomize/releases/tag/kustomize%2Fv3.8.6
./kustomize build  tf-job-crds/overlays/application | kubectl apply -f  -
./kustomize build tf-job-operator/overlays/application  | kubectl  apply -f  -

4.查看pod是否running

kubectl  get   all  -n  kubeflow