Bug ID | 1105421 |
---|---|
Summary | CoreDNS doesn't start |
Classification | openSUSE |
Product | openSUSE Tumbleweed |
Version | Current |
Hardware | Other |
OS | Other |
Status | NEW |
Severity | Normal |
Priority | P5 - None |
Component | Kubic |
Assignee | kubic-bugs@opensuse.org |
Reporter | pgeorgiadis@suse.com |
QA Contact | qa-bugs@suse.de |
Found By | --- |
Blocker | --- |
I followed the guide at https://kubic.opensuse.org/blog/2018-08-20-kubeadm-intro/ As I have previously said [1], the CNI plugin binaries are misplaced, thus the pods cannot talk to each other. The kubeadm cluster is able to join nodes (you can happily see them: kubectl get nodes) but it's kinda useless the cluster cannot deploy any containers or kube-apps on it. [1] https://bugzilla.novell.com/show_bug.cgi?id=1093132#c29 Feel free to re-open: https://github.com/kubernetes/kubernetes/issues/67390 Reproducer: - Follow https://kubic.opensuse.org/blog/2018-08-20-kubeadm-intro/ - At the end, do: kubectl get pods --namespace=kube-system you will notice two pods of core-dns 0/1 with status "ContainerCreating". This stays forever, thus there's no DNS in the cluster. This happens because kubelet looks at Ubuntu's location by default for CNI plugins: > # cat /var/lib/kubelet/kubeadm-flags.env > KUBELET_KUBEADM_ARGS=--cgroup-driver=cgroupfs --cni-bin-dir=/opt/cni/bin --cni-conf-dir=/etc/cni/net.d --network-plugin=cni Workaround: ~~~~~~~~~~~ Move the cni-plugins into /opt/cni/bin (not recommended) find a way to tell kubelet to look into the correct path (recommended) or even better, use CRI-O instead of Docker and you won't have that problem. See the steps: https://pastebin.com/raw/bJB0zJLD