新添加节点containerd服务没有读取/etc/containerd/config.toml里sandbox_image，而是依然下载registry.k8s.io/pause:3.8

[asianuxchina]

openeuler 24.03 sp1

kubekey 3.1.9

1、containerd的配置/etc/containerd/config.toml里sandbox_image已经替换为registry.aliyuncs.com/google_containers/pause:3.9。

2、执行新添加节点执行过程没有报错，但是新添加节点pod一直处于pending状态。

3、排查新节点containerd服务报level=error msg="RunPodSandbox for &PodSandboxMetadata{Name:haproxy-worker-2,Namespace:kube-system,Attempt:0,} failed, error" error="failed to get sandbox image "registry.k8s.io/pause:3.8": failed to pull image "registry.k8s.io/pause:3.8": failed to pull and unpack image "registry.k8s.io/pause:3.8": failed to resolve reference "registry.k8s.io/pause:3.8": failed to do request: Head "[https://registry.k8s.io/v2/pause/manifests/3.8](https://registry.k8s.io/v2/pause/manifests/3.8%5C)": dial tcp [::]:443: i/o timeout"
containerd服务加载默认配置去下载registry.k8s.io/pause:3.8，没有读取/etc/containerd/config.toml里定义sandbox_image=registry.aliyuncs.com/google_containers/pause:3.9，执行systemctl restart containerd后能够恢复，containerd读取/etc/containerd/config.toml配置，pod能正常启动。

4、能否在执行添加节点后增加一次containerd服务重启操作？在哪里配置能够实现添加节点后containerd强制执行一次重启服务加载/etc/containerd/config.toml。

[skyhhjmk]

尝试设置环境变量以让kubekey使用镜像源

export KKZONE=cn

[redscholar]

@asianuxchina 是新增的节点，已经安装过了containerd么。