grpc connection is unavailable

[kirankt]

grpc connection is unavailable

[kirankt]

Also, I'm using OCP11:

[root@kt-c7kb9 contrib]# oc version
oc v3.11.0-alpha.0+e0f1909-460
kubernetes v1.11.0+d4cacc0
features: Basic-Auth GSSAPI Kerberos SPNEGO

Server https://127.0.0.1:8443
openshift v3.11.0-alpha.0+bb11f51-481
kubernetes v1.11.0+d4cacc0

[Akrog]

This is a communication issue between kubelet and the Node side of the CSI plugin. Kubelet cannot establish the GRPC connection with the CSI plugin to call NodeStageVolume.

I can think of a couple of reasons why this can happen:

• Wrong provisioner set somewhere: Check StorageClass and the provisioner parameter passed to the external provisioner.
• Wrong socket path: Check the socket path of the Node CSI plugin container or the CSI container if running a Controller+Node deployment.

[kirankt]

Hi,
Thanks for the suggestion. As you pointed out, everything looks great when the CSI_MODE is 'all'. When they are separated, I don't see any activity in the driver while using "node" mode. Let me dig a little further to provide you more details.

[kirankt]

I've gone past this issue. On OCP, the CSI driver should host mount /var/lib/origin/openshift.local.volumes rather than /var/lib/kubelet. Interesting to note that the CSI socket directory still resides in /var/lib/kubevirt e.g. /var/lib/kubelet/plugins/com.redhat.cinderlib-csi

[Akrog]

Thank you very much for the info, and that is certainly odd, I'll have a look and see where does that difference come from.

If you are splitting the plugin into controller and node services, you may be interested in looking into the manifests I have in the new Kubernetes demo (in the kubeyml directory) that are splitting the services: https://github.com/Akrog/cinderlib-csi/tree/master/examples/kubernetes