- All nodes within a Platform9 Managed Kubernetes (PMK) cluster appear to be in a "NotReady" state.
- In the PMK UI, the nodes may appear to similarly be in a "failed" state.
- In the pf9-kube service log on the respective master node(s), the following error is observed.
-- /var/log/pf9/kube/kube.log --
Keepalived_vrrp: Cant find interface bond0 for vrrp_instance K8S_APISERVER !!!
- Platform9 Managed Kubernetes - All Versions
The interface which was specified for the VIP – bond0, for example – on the master node no longer exists.
- Identify the VIP interface provided during the time of configuration as defined in /etc/pf9/kube.env on the master node.
- Ensure that the specified interface exists.
- If the interface does not exist – configure it, and then reboot the node (especially to ensure that its configuration persists).
Note: If it is desired to modify the VIP interface, it cannot be patched via the Qbert (cluster-management) API and, thus, unfortunately, you must otherwise re-create the cluster.