This page documents the list of known issues and possible work arounds/solutions.
Problem
OIDC user is denied to access the user cluster in the KKP with K8s version of 1.20 and below. Refer the github issue Bug: OIDC authentication… for detailed problem description. Example logs look like below,
Kubectl output
kubectl get nodes
error: You must be logged in to the server (Unauthorized)
API server logs
2022-05-26T11:46:11.269134597Z stderr F E0526 11:46:11.267368 1 authentication.go:63] "Unable to authenticate the request" err="[invalid bearer token, oidc: authenticator not initialized]"
2022-05-26 13:46:11
2022-05-26T11:46:11.200645694Z stderr F E0526 11:46:11.200494 1 authentication.go:63] "Unable to authenticate the request" err="[invalid bearer token, oidc: authenticator not initialized]"
2022-05-26 13:46:10
2022-05-26T11:46:10.282230799Z stderr F E0526 11:46:10.282080 1 oidc.go:224] oidc authenticator: initializing plugin: Get "https://<your-kkp.domain>/dex/.well-known/openid-configuration": net/http: request canceled while waiting for connection (Client.Timeout exceeded while awaiting headers)
Root cause
The KKP API Server network policy is relying on the namespace label kubernetes.io/metadata.name which is automatically present on K8s 1.21 and above versions, but missing on K8s versions below 1.21. Due to the mismatch in the label selector, the access is denied.
Solution
As the issue is seen only with older versions of K8s which have reached end of life, the preferred solution is to upgrade the K8s to 1.21 or the latest version.
In the case where upgrade is not desirable then a work around can be applied by adding a label to the nginx-ingress-controller namespace as shown below.
kubectl label ns nginx-ingress-controller "kubernetes.io/metadata.name=nginx-ingress-controller"
Problem
In a KKP user cluster with Cilium CNI and IPVS kube-proxy mode, the connectivity between the NodePort service and client pod does not work when the service is load balanced to a pod running on a remote node. For the detailed description and the steps to reproduce the problem, refer issue #8767.
Root Cause
IPVS kube-proxy mode is not really supported by Cilium as mentioned in the Cilium issue #18610.
Solution
We do not recommend to configure the Cilium with IPVS kube-proxy mode and this option has been removed from the KKP UI as part of the issue #4687.
Problem
The issue arises in Ubuntu 22.04 cloud image OVAs starting from version 20230602 when they are run on VMware Cloud Director. This problem disrupts the provisioning of new Kubernetes nodes using machine-controller due to interruptions caused by reboots.
Root Cause
The root cause of this issue can be traced back to a change in the default settings of open-vm-tools. These changes, in turn, affect the behavior of cloud-init during startup, leading to the disruptive behavior observed when provisioning new Kubernetes nodes. Specifically, the open-vm-tools.service starts before cloud-init, and it runs with the default timeout (30 seconds).
Solution
One interim solution in this scenario is to create a custom Ubuntu 22.04 image with the following setting preconfigured in /etc/vmware-tools/tools.conf file.
[deployPkg]
wait-cloudinit-timeout=0
This adjustment will help ensure that the issue no longer disrupts the provisioning of new Kubernetes nodes on the affected Ubuntu 22.04 cloud images running on VMware Cloud Director provider.
For additional details and discussions related to this issue, you can refer to the following GitHub issues: