polardbx-operator install MountVolume.SetUp failed for volume "etclocaltime" : hostPath type check failed: /etc/localtime is not a file

[wyzhl126]

环境单机docker安装的rancher单节点 上安装
[root@test]# kubectl -n polardbx get pods
NAME READY STATUS RESTARTS AGE
polardbx-c00-gt96-dn-0-cand-0 0/3 ContainerCreating 0 24m
polardbx-c00-gt96-dn-0-cand-1 0/3 ContainerCreating 0 24m
polardbx-c00-gt96-dn-0-log-0 0/3 ContainerCreating 0 24m
polardbx-c00-gt96-gms-cand-0 0/3 ContainerCreating 0 24m
polardbx-c00-gt96-gms-cand-1 0/3 ContainerCreating 0 24m
polardbx-c00-gt96-gms-log-0 0/3 ContainerCreating 0 24m
polardbx-controller-manager-7b4f47dfcf-5txkz 1/1 Running 0 38m
polardbx-hpfs-2fq8x 1/1 Running 0 38m
polardbx-tools-updater-jd829 1/1 Running 0 38m

kubectl -n polardbx describe pod polardbx-c00-gt96-dn-0-cand-0
Events:
Type Reason Age From Message

---

Normal Scheduled 17s default-scheduler Successfully assigned polardbx/polardbx-c00-gt96-dn-0-cand-0 to dz-fanjie
Warning FailedMount 2s (x6 over 17s) kubelet MountVolume.SetUp failed for volume "etclocaltime" : hostPath type check failed: /etc/localtime is not a file
Warning FailedMount 2s (x6 over 17s) kubelet MountVolume.SetUp failed for volume "zoneinfo" : hostPath type check failed: /usr/share/zoneinfo is not a directory

出现上述错误，宿主机及docker容器都有上面的2个目录。

[Vettal]

@wyzhl126 It seems to be a result of the containerized kubelet process in RKE and Rancher launched Kubernetes clusters. In your host, the kubelet is running in a container, and the path /etc/localtime , /usr/share/zoneinfo(mount type is Directory instead DirectoryOrCreate) dose not be mounted in the kubelet container. So the hostPath type check failed.

Following are some related issues:
rancher/rancher#22882
rancher/rancher#14836

Workaround:

Edit your cluster config in rancher cluster management page, add extra_binds into kubelet service as the following:

services:
  kubelet:
      extra_binds:
        - '/etc/localtime:/etc/localtime'
        - '/usr/share/zoneinfo:/usr/share/zoneinfo'