How to get started, and accomplish tasks, using Kubernetes.

Edit This Page

VMware vSphere

This page covers how to get started with deploying Kubernetes on vSphere and details for how to configure the vSphere Cloud Provider.

Getting started with vSphere

Kubernetes comes with a cloud provider for vSphere. A quick and easy way to try out the cloud provider is to deploy Kubernetes using Kubernetes-Anywhere.

This page also describes how to configure and get started with the cloud provider if deploying using custom install scripts.

Deploy Kubernetes on vSphere

To start using Kubernetes on top of vSphere and use the vSphere Cloud Provider use Kubernetes-Anywhere. Kubernetes-Anywhere will deploy and configure a cluster from scratch.

Detailed steps can be found at the getting started with Kubernetes-Anywhere on vSphere page

vSphere Cloud Provider

vSphere Cloud Provider allows using vSphere managed storage within Kubernetes. It supports:

  1. Volumes
  2. Persistent Volumes
  3. Storage Classes and provisioning of volumes.

Documentation for how to use vSphere managed storage can be found in the persistent volumes user guide and the volumes user guide

Examples can be found here

Configuring vSphere Cloud Provider

If a Kubernetes cluster has not been deployed using Kubernetes-Anywhere, follow the instructions below to use the vSphere Cloud Provider. These steps are not needed when using Kubernetes-Anywhere, they will be done as part of the deployment.

This can be done via govc tool

export GOVC_URL=<IP/URL>
export GOVC_USERNAME=<vCenter User>
export GOVC_PASSWORD=<vCenter Password>
govc vm.change -e="disk.enableUUID=1" -vm=<VMNAME>

Sample Config:

        user = <User name for vCenter>
        password = <Password for vCenter>
        server = <IP/URL for vCenter>
        port = <Default 443 for vCenter>
        insecure-flag = <set to 1 if the host above uses a self-signed cert>
        datacenter = <Datacenter to be used>
        datastore = <Datastore to use for provisioning volumes using storage classes/dynamic provisioning>
        working-dir = <Folder in which VMs are provisioned, can be null>
	scsicontrollertype = pvscsi

Known issues

Kube-up (Deprecated) is no longer supported and is deprecated. The steps for kube-up are included but going forward kube-anywhere is preferred.

The recommended version for kube-up is v1.4.7

The example below creates a Kubernetes cluster with 4 worker node Virtual. Machines and a master Virtual Machine (i.e. 5 VMs in your cluster). This cluster is set up and controlled from your workstation (or wherever you find convenient).


export GOPATH=$HOME/src/go
mkdir -p $GOPATH
export PATH=$PATH:$GOPATH/bin
# Sample commands for v0.8.0 for 64 bit Linux.
curl -OL
gzip -d govc_linux_amd64.gz
chmod +x govc_linux_amd64
mv govc_linux_amd64 /usr/local/bin/govc


Download a prebuilt Debian 8.2 VMDK that we’ll use as a base image:

curl --remote-name-all{,.md5}
md5sum -c kube.vmdk.gz.md5
gzip -d kube.vmdk.gz

Configure the environment for govc

export GOVC_URL='hostname' # hostname of the vc
export GOVC_USERNAME='username' # username for logging into the vsphere.
export GOVC_PASSWORD='password' # password for the above username
export GOVC_NETWORK='Network Name' # Name of the network the vms should join. Many times it could be "VM Network"
export GOVC_INSECURE=1 # If the host above uses a self-signed cert
export GOVC_DATASTORE='target datastore'
# To get resource pool via govc: govc ls -l 'host/*' | grep ResourcePool | awk '{print $1}' | xargs -n1 -t govc
export GOVC_RESOURCE_POOL='resource pool or cluster with access to datastore'
export GOVC_GUEST_LOGIN='kube:kube' # Used for logging into kube.vmdk during deployment.
export GOVC_PORT=443 # The port to be used by vSphere cloud provider plugin
# To get datacente via govc: govc
export GOVC_DATACENTER='ha-datacenter' # The datacenter to be used by vSphere cloud provider plugin

Sample environment

export GOVC_URL=''
export GOVC_USERNAME='administrator'
export GOVC_PASSWORD='MyPassword1'
export GOVC_NETWORK='VM Network'
export GOVC_DATASTORE='datastore1'
export GOVC_RESOURCE_POOL='/Datacenter/host/'
export GOVC_GUEST_LOGIN='kube:kube'
export GOVC_PORT='443'
export GOVC_DATACENTER='Datacenter'

Import this VMDK into your vSphere datastore:

govc import.vmdk kube.vmdk ./kube/

Verify that the VMDK was correctly uploaded and expanded to ~3GiB:

govc ./kube/

If you need to debug any part of the deployment, the guest login for the image that you imported is kube:kube. It is normally specified in the GOVC_GUEST_LOGIN parameter above.

Also take a look at the file cluster/vsphere/ and make any needed changes. You can configure the number of nodes as well as the IP subnets you have made available to Kubernetes, pods, and services.

Starting a cluster

Now, let’s continue with deploying Kubernetes. This process takes about ~20-30 minutes depending on your network.

From extracted binary release
cd kubernetes
KUBERNETES_PROVIDER=vsphere cluster/
Build from source
cd kubernetes
make release
KUBERNETES_PROVIDER=vsphere cluster/

Refer to the top level README and the getting started guide for Google Compute Engine. Once you have successfully reached this point, your vSphere Kubernetes deployment works just as any other one!


Extra: debugging deployment failure

The output of displays the IP addresses of the VMs it deploys. You can log into any VM as the kube user to poke around and figure out what is going on (find yourself authorized with your SSH key, or use the password kube otherwise).

Support Level

IaaS Provider Config. Mgmt OS Networking Docs Conforms Support Level
Vmware vSphere Kube-anywhere Photon OS Flannel docs   Community (@abrarshivani), (@kerneltime), (@BaluDontu)(@luomiao)

For support level information on all solutions, see the Table of solutions chart.

Analytics Create an Issue Edit this Page