如何解决ks-console:v3.4.1被转移, 导致kubesphere console 无法正常启动?

jerryhou168

最近发现 kubesphere console 打不开, 原因是这三个image已经从kubesphere 被转移到kubespheredev下面了

请问如何修改现在的环境，使得kubesphere console 可以继续正常工作？谢谢

Type Reason Age From Message

Normal BackOff 2m48s (x1125 over 4h22m) kubelet Back-off pulling image “kubesphere/ks-console:v3.4.1”

Type Reason Age From Message

Warning Failed 33m (x50 over 4h26m) kubelet Failed to pull image “kubesphere/ks-controller-manager:v3.4.1”: rpc error: code = Unknown desc = Error response from daemon: manifest for kubesphere/ks-controller-manager:v3.4.1 not found: manifest unknown: manifest unknown

Normal Pulling 28m (x51 over 4h26m) kubelet Pulling image “kubesphere/ks-controller-manager:v3.4.1”

Normal BackOff 3m19s (x1131 over 4h26m) kubelet Back-off pulling image “kubesphere/ks-controller-manager:v3.4.1”

Type Reason Age From Message

Normal BackOff 3m44s (x1125 over 4h23m) kubelet Back-off pulling image “kubesphere/ks-console:v3.4.1”

cici

@jerryhou168 把deployment replica先設0讓pod停掉再設定image然後把replica設回1 pod會自動啟動試看看

cici

namespace kubesphere-system這幾個應該都是同樣的閉源官方image這幾個都被拿掉了==

如果你沒自己image庫是從官方拉就會有這問題，這namespace幾個的解法都一樣可試看看

jerryhou168

cici 谢谢cici, 我们也是这样解决的，但是每次VM 重启之后就需要再次做相同的修改.

echo “patch image for ks-console”

kubectl -n kubesphere-system set image deployment/ks-controller-manager ks-controller-manager=kubespheredev/ks-controller-manager:v3.4.1

kubectl -n kubesphere-system set image deployment/ks-apiserver ks-apiserver=kubespheredev/ks-apiserver:v3.4.1

kubectl -n kubesphere-system set image deployment/ks-console ks-console=kubespheredev/ks-console:v3.4.1

echo “patch image pulling policy”

kubectl -n kubesphere-system patch deployment ks-controller-manager –type=‘json’ -p=‘[{“op”:“replace”,“path”:“/spec/template/spec/containers/0/imagePullPolicy”,“value”:“IfNotPresent”}]’

kubectl -n kubesphere-system patch deployment ks-apiserver –type=‘json’ -p=‘[{“op”:“replace”,“path”:“/spec/template/spec/containers/0/imagePullPolicy”,“value”:“IfNotPresent”}]’

kubectl -n kubesphere-system patch deployment ks-console –type=‘json’ -p=‘[{“op”:“replace”,“path”:“/spec/template/spec/containers/0/imagePullPolicy”,“value”:“IfNotPresent”}]’

echo “restart”

kubectl -n kubesphere-system rollout restart deployment ks-controller-manager

kubectl -n kubesphere-system rollout restart deployment ks-apiserver

kubectl -n kubesphere-system rollout restart deployment ks-console

zhuyifei-ruichuang

我在测试环境临时处理的方法如下，

kubesphere console 部署在master node。部署时会拉取镜像，假设你的runtime是docker，可查询master node是否有该镜像，例如，

root@master1:/opt/kubesphere# docker images | grep ks-console
kubesphere/ks-console                         v3.4.1                         89910c20821f   21 months ago   146MB

确认镜像在本地存在，找pod的配置，先找到POD，

root@master1:/opt/kubesphere# kubectl get pod -A -o wide | grep console
kubesphere-system              ks-console-7f8cf79d65-8pm9j                                    1/1     Running            1 (2d1h ago)      2d5h   10.233.97.14    master1   <none>           <none>

找POD的控制器，

root@master1:/opt/kubesphere# kubectl describe -n kubesphere-system pods/ks-console-7f8cf79d65-8pm9j | grep Contro
Controlled By:  ReplicaSet/ks-console-7f8cf79d65

编辑控制器，

root@master1:/opt/kubesphere# kubectl edit -n kubesphere-system ReplicaSet/ks-console-7f8cf79d65

找到 imagePullPolicy,修改为 imagePullPolicy: IfNotPresent，查看POD状态，可见POD恢复正常。若未恢复正常，就delete pod，等待自动创建POD，是delete POD，不是删除控制器。

希望可以帮到你。

其他异常的POD都是这个方法进行处理，若node上没有镜像，就在有镜像的node导出镜像到所有node。

zhuyifei-ruichuang

需补充一下，如果是测试，这个方法还是可以参考的，可用但难以保障可靠。如果后续明确计划商用，还是直接内部商议用商业版+延长试用期的方式，直接上商用。

zhuyifei-ruichuang