KubeOne(1)

📦 version 1.12.3, in arch rolling

💅 kubeone-migrate-to-ccm-csi — KubeOne(1) ✨

okay so bestie, we're looking at the KubeOne(1) manual and honestly? this one's a glow-up arc for your cluster. sit down, grab a lil drink, let's get into it 🍓

🌸 NAME 🌸

kubeone-migrate-to-ccm-csi — moves your live cluster off the in-tree cloud provider and onto the external cloud-controller-manager (CCM) plus the CSI plugin. like, a whole living-cluster transplant, no cap 💀

🦋 SYNOPSIS 🦋

kubeone migrate to-ccm-csi [flags]

✨ DESCRIPTION ✨

so the in-tree cloud provider is getting deprecated (http://kep.k8s.io/2395), which is giving "we're so over you" energy, and this command is literally the designated exit strategy. it helps you migrate existing clusters from the in-tree cloud provider over to the external cloud-controller-manager (CCM) and CSI plugin. iconic behavior only 👑

real talk though: if your cluster was created with .cloudProvider.external enabled, the CCM/CSI migration is not needed because the cluster is already using external CCM. you already ate, no notes 💅

migration is currently available for OpenStack and vSphere. other providers will be added in future KubeOne releases — patience, bestie 🧚

the migration is done in two phases:

  • Phase 1: deploy external CCM and CSI plugin, while leaving in-tree provider enabled. Kubernetes API server and kube-controller-manager are configured to:
    • use controllers integrated in external CCM instead of in-tree cloud provider for all cloud-related operations
    • redirect all volumes-related operations to the CSI plugin

the existing worker nodes will continue to use in-tree provider (that's why it's still left enabled), so therefore, all worker nodes managed by machine-controller must be rolled out after phase 1 is complete. yes, all of them. don't skip 💀

  • Phase 2: complete the CCM/CSI migration by fully-disabling in-tree provider. to trigger the phase 2, users need to run "kubeone migrate to-ccm-csi" command with the --complete flag. this should be done after all worker nodes managed by machine-controller are rolled-out. periodt.

make sure to familiarize yourself with the CCM/CSI migration requirements by checking the following document: https://docs.kubermatic.com/kubeone/main/guides/ccm-csi-migration/ — please actually read it, i'm begging 🙈

🎀 OPTIONS 🎀

  • -y, --auto-approve[=false] — auto approve plan. no lil "are you sure?" moment, she just goes 🚀
  • --complete[=false] — complete ccm/csi migration. this is the phase 2 trigger, the main character of the flags ⭐
  • -h, --help[=false] — help for to-ccm-csi. classic, reliable, always there for you 🫶

🌈 OPTIONS INHERITED FROM PARENT COMMANDS 🌈

  • -c, --credentials="" — File to source credentials and secrets from 🔐
  • -d, --debug[=false] — debug output with stacktrace. for when you need the whole lore, uncut 😭
  • -l, --log-format="text" — format for logging 📝
  • -m, --manifest="./kubeone.yaml" — Path to the KubeOne config. the sacred text fr fr 📜
  • -t, --tfjson="" — Source for terraform output in JSON - to read from stdin. If path is a file, contents will be used. If path is a dictionary, terraform output -json is executed in this path 🧚
  • -v, --verbose[=false] — verbose output. she talks more. we love her for it 💖

👀 SEE ALSO 👀

🌸 HISTORY 🌸

8-Feb-2026 — Auto generated by spf13/cobra 🤖 (yes the robots wrote the receipts, we just added the sparkle)

okay that's the whole thing, bestie, you're fully caught up 🎀 ngl this page was Auto generated by spf13/cobra back in Feb 2026, and that's the KubeOne(1) manual wrapped with a bow — go migrate something and slay ✨💅🦋

📦 where this page comes from