Hi,
sorry to pick up this old post, but I faced the very same issue on multiple ocassions, but I remembered that I had this issue before and how I fixed it. But the questions still remains. I have a fresh kolla-ansible yoga magnum deployment with latest Fedora CoreOS 35 image and I always fail to deploy a K8s cluster:
kubectl get pods -A
kube-system csi-cinder-controllerplugin-0 4/5 CrashLoopBackOff 6 (4m54s ago) 14m
I can fix it by specifying a label of:
csi_snapshot_tag=v4.0.1
But this only works If I use a local registry where I download
plenty of container images from multipe sources.
It failed when not using a local registry since magnum tries to download the csi_snapshotter docker image from a different source:
Events:
Type Reason Age
From Message
---- ------ ----
---- -------
Normal Scheduled 13m
default-scheduler Successfully assigned
kube-system/csi-cinder-controllerplugin-0 to
k8s-flannel-small-35-mi6fcba4mpdc-master-0
Warning NetworkNotReady 13m (x24 over 13m)
kubelet network is not ready: container runtime network
not ready: NetworkReady=false reason:NetworkPluginNotReady
message:docker: network plugin is not ready: cni config
uninitialized
Normal Pulling 13m
kubelet Pulling image
"quay.io/k8scsi/csi-attacher:v2.0.0"
Normal BackOff 3m43s (x33 over 11m)
kubelet Back-off pulling image
"quay.io/k8scsi/csi-snapshotter:v4.0.1"
BTW for the local registry, It seems that the list of images needed described here:
https://docs.openstack.org/magnum/yoga/user/
is incomplete and outdated. I had to pull a couple more. No complaint, just mentioning it. :)
I wonder if the issue is known and if it is if there is note for it in the release notes for magnum under yoga?
I have not tried ZED yet since my OS is Rocky Linux 8.x and there
seems to be now upgrade path available yet.
Cheers and a nice weekend
Oliver
[kolla-ansible] [yoga] [magnum] [k8s]