DEV Community: Rajesh N

FluxCD Helm Repository - chart pull error: 401

Rajesh N — Fri, 21 Mar 2025 02:05:12 +0000

Objective: Many of us used Flux CD helm repository and helm release with open repository on which anonymous access was allowed, but when we are using the authentication-enabled repository, it's difficult to fix helm release.

In my organization, we use JFrog Artifactory and we enabled authentication for all the repos. So I prepared the helm release and helm repository with an authentication-enabled method.

Simple helm repository with credential got this error

chart pull error: failed to download chart for remote reference: failed to fetch https://repo.ops.homelab.com/artifactory/dcops-helm-repo/keda-2.16.1.tgz : 401

apiVersion: source.toolkit.fluxcd.io/v1
kind: HelmRepository
metadata:
name: dcops-helm-repo
namespace: flux-system
spec:
interval: 15m
provider: generic
passCredentials: true #fix
url: "https://repo.ops.homelab.com/artifactory/helm-repo"
secretRef:
name: artifact-credential

How to control the cluster nodes recreation in Tanzu

Rajesh N — Mon, 28 Oct 2024 05:18:16 +0000

In a Tanzu Kubernetes cluster, the Supervisor cluster continuously monitors and reconciles the cluster. If any worker node fails to respond within the monitoring interval, it is marked as failed, and a new worker node is created to replace it. However, this operation may not always be accurate. If you plan to perform any maintenance on the Tanzu environment, it is necessary to pause the cluster reconciliation; otherwise, you may see many nodes being recreated.

If you are planning any network maintenance within your organization and expect any storage latency or network latency/interruption, it is better to pause the cluster reconciliation.

How to Pause the Cluster Reconciliation

Login to the Supervisor Node:
root@vcenter [ ~ ]# sshpass -p /usr/lib/vmware-wcp/decryptK8Pwd.py |grep PWD |awk '{print $2}' ssh -o StrictHostKeyChecking=no root@/usr/lib/vmware-wcp/decryptK8Pwd.py |grep IP |awk '{print $2}'

To Pause the Cluster Reconciliation:
kubectl patch cluster --type merge -p '{"spec":{"paused": true}}' -n tanzu-vns tanzu-cluster

To Resume the Cluster Reconciliation:
kubectl patch cluster --type merge -p '{"spec":{"paused": false}}' -n tanzu-vns tanzu-cluster

tanzu worker node created with wrong operating system

Rajesh N — Mon, 28 Oct 2024 05:04:55 +0000

In the VMware 8 Tanzu environment, after upgrading to WCP 1.27, worker nodes and control plane nodes are created with the default operating system (Photon) even though Ubuntu was selected in the TKR selection.

This issue was observed only in existing Tanzu Kubernetes clusters. When the worker nodes are created with the wrong operating system, some system pods may remain in a pending state due to node selector labels not matching on the newly created worker nodes.

Workaround: Apply the node label with the Ubuntu version tag:

kubectl label node --overwrite run.tanzu.vmware.com/kubernetesDistributionVersion=v1.25.13---vmware.1-fips.1-tkg.1.ubuntu

To apply the label to all nodes:

kubectl get nodes | awk '{ print $1 }' | xargs -I {} kubectl label nodes {} run.tanzu.vmware.com/kubernetesDistributionVersion=v1.25.13---vmware.1-fips.1-tkg.1.ubuntu --overwrite

In vSphere version 8u3, Tanzu introduced a TKC annotation for the operating system:

run.tanzu.vmware.com/resolve-os-image: os-name=ubuntu

We need to annotate the cluster with run.tanzu.vmware.com/resolve-os-image:

kubectl annotate tkc -n kubernetes-vns run.tanzu.vmware.com/resolve-os-image="os-name=ubuntu"