0
I'm trying to configure the Cinder CSI plugin (https://github.com/kubernetes/cloud-provider-openstack/blob/master/docs/using-cinder-csi-plugin.md) on my Kubernetes cluster (installed on OpenStack instances).
Everything works fine (the PV, PVC and actual Cinder volume are created just fine) but when the csi-attacher
tries to attach my Cinder volume to my node, it fails with:
Error processing "csi-0c1aee9c7e99bc90a586541a4bb7974e38978dc3a2c441a1ad1413d7435d886c": failed to attach: rpc error: code = Unknown desc = failed to attach 445d9931-9e32-48b5-b765-209a4526691f volume to i-0031336c compute: Resource not found
Note that the volume ID is right though:
> cinder list | grep 445d9931-9e32-48b5-b765-209a4526691f
| 445d9931-9e32-48b5-b765-209a4526691f | available | pvc-788eb00a-99bc-11e9-b7dc-fa163eb3415c | 1 | classic | false | |
My node is not i-0031336c
but an actual UUID:
> nova list | grep worker03-dev
| 9d7cd2bb-a365-4276-86ed-f4884132c361 | worker03-dev | ACTIVE | - | Running | Ext-Net=<redacted> |
I don't know why, but something (I'm not yet quite expert on the CSI side) is adding an annotation with the wrong ID to my nodes:
> kubectl get node worker03-dev.node.consul -o yaml | grep nodeid
csi.volume.kubernetes.io/nodeid: '{"cinder.csi.openstack.org":"i-0031336c"}'
When querying the OpenStack API from the host, it returns the right UUID too:
root@worker03-dev ~ # curl -s 169.254.169.254/openstack/latest/meta_data.json | jq '.uuid'
"9d7cd2bb-a365-4276-86ed-f4884132c361"
Do I have something completely wrong?
I run these CSI containers:
image: quay.io/k8scsi/csi-attacher:v1.1.1
image: quay.io/k8scsi/csi-provisioner:v1.0.1
image: quay.io/k8scsi/csi-snapshotter:v1.0.1
image: quay.io/k8scsi/csi-node-driver-registrar:v1.1.0
image: docker.io/k8scloudprovider/cinder-csi-plugin:latest
(I started with v1.0.1
for all k8scsi
containers but tried to upgrade some of them without success)