Cannot run hostpath integration example about docs HOT 9 CLOSED

I ended up fixing the problem with the following script:

#!/bin/bash
minikube start --vm-driver=hyperkit --kubernetes-version v1.10.0 --feature-gates=KubeletPluginsWatcher=true,BlockVolume=true,CSIBlockVolume=true --bootstrapper=localkube

# replace /var sys links with abs links
VAR_CMD="sudo find /var -type l -execdir bash -c 'ln -sfn \"\$(readlink -f \"\$0\")\" \"\$0\"' {} \\;"
minikube ssh "$VAR_CMD"

# replace etc sys links with abs links
ETC_CMD="sudo find /etc -type l -execdir bash -c 'ln -sfn \"\$(readlink -f \"\$0\")\" \"\$0\"' {} \\;"
minikube ssh "$ETC_CMD"

The syslinks from /run to /var/run where misbehaving when launching pods with bind mounts due to them being relative syslinks. Changing them to absolute syslinks using the last two commands fixed that. Kubeadm gave me a 80-20 split of not working/working so I switched to localkube and everything seems to be working now!

Hope this helps someone and prevents them from delving down into this particular rabbit hold.

from docs.

Upon further inspection it seems that once the host-path driver container starts, the docker daemon cleans up the /var/run directory. /var/run/docker.sock disappears alongside a variety of other files in /var/run/docker, although systemctl still shows the daemon as active and running. The log file for the hostpath-driver container is empty as well. I have tried this with a few other csi-drivers such as the mock-driver and the same behavior occurs.

from docs.

Further information: this seems to only happen when I run it using minikube. I ran it on normal cluster and it seems to be working fine. So there must be some issue with how I am setting up my minikube cluster

from docs.

I think I faced the same issue and was able to create an easy reproducer for it. I filed a minikube issue
kubernetes/minikube#3378

from docs.

@oleksiys @sean-dooher: thanks for your analysis on this one. Do you agree that we can close this issue now, as it seems to be caused by minikube?

We could of course document the issue, but I don't think we should - that would become unmanageable quickly.

from docs.

I agree, I think we can close this issue and add a reference to the minikube issue from the "Known issues" section.

from docs.

from docs.

Above command doesn't work with Kubernetes version 1.12.4.

This PR shows localkube removed from latest minikube.
kubernetes/minikube#2911

As per csi documentation, steps are written for Kubernetes 1.12 version.

from docs.

minikube 1.1.1 & Kubernetes v1.14.3 have same issue》

from docs.