[BUG] MountVolume.SetUp failed in GKE

[valiantljk]

What is your environment(Kubernetes version, Fluid version, etc.)
Kubernetes: 1.15
Fluid: 0.3.0
Helm: v3
Describe the bug
Followed instructions at here
Trying to install fluid with Helm3 on GKE. Found that PODS in fluid-system stuck in 'container creating' status. Only controller-manager pod runs well. Here is the log:
kubectl describe pod/csi-nodeplugin-fluid-6dsff -n fluid-system MountVolume.SetUp failed for volume "fluid-src-dir" : mkdir /alluxio-mnt: read-only file system

After changing the csi-nodeplugin-fluid daemonset.yaml by specifying the MountVolume as readOnly: False,
still not working.
What you expect to happen:
Fluid can be deployed on GKE
How to reproduce it

Helm install fluid fluid/

Additional Information

[cheyang]

Can you run mkdir /alluxio-mnt in the k8s node?

You can use the follow command to check the node info.

kubectl get po -n fluid-system -o wide

[cheyang]

Does GKE's kubelet has authority to create directory in k8s node?

[valiantljk]

It looks like GKE node's root file system is read only.

Does alluxio-mnt has to be in root / ?

[cheyang]

I've fixed it in #223. And please delete the existing chart and install with the chart: https://github.com/fluid-cloudnative/fluid/blob/fb123c104c4e0d8707260fcbb2b6de32f868054a/charts/fluid/fluid/values.yaml

[cheyang]

And please follow the updated guide: https://github.com/fluid-cloudnative/fluid/blob/master/docs/zh/samples/accelerate_data_accessing.md

And notice:

注意: 上述mountPoint中使用了Apache清华镜像源进行演示，如果你的环境位于海外，请更换为https://downloads.apache.org/hbase/2.2.5/进行尝试