]

Solution

  • Modern Applications

Type

  • Document

Level

  • Intermediate

Category

  • Quick-Start

Product

  • vSphere
  • vSphere 7
  • vSphere with Tanzu

Technology

  • Kubernetes

Phase

  • Deploy

vSphere with Tanzu Quick Start Guide V1a

Introduction

vSphere with Tanzu is the latest update to Kubernetes running natively on vSphere. The biggest change with vSphere with Tanzu is that introduces the ability to enable Kubernetes on vSphere clusters using a vSphere Distributed Switch.

vSphere with Tanzu utilizes vSphere Distributed Switch Portgroups and a “bring your own” network strategy for load balancing Kubernetes workloads. The initial release supports HAProxy for load balancing via our new Load Balancer API. Look for additional load balancers coming soon.

Scope of this Document

The guiding principle of this document is to get to a working evaluation of vSphere with Tanzu. You can create this environment on physical hardware or via nested virtual machines. You can also do everything in this document within your standard VMware evaluation licensing window.

Setting up and installing vSphere with Tanzu, regardless of using NSX or vSphere Distributed Switch requires custom networking configuration depending on your environment. Because so many customer’s configurations are unique it’s very difficult to test every configuration.

With that in mind and to ensure you can get vSphere with Tanzu up and running as quickly as possible on an evaluation basis, we have limited the networking scope of this guide to using one subnet for workloads and virtual IP’s (VIP) and one subnet for vSphere management components (vCenter, ESXi).

Note: This is NOT a replacement for the documentation, nor should this configuration be used in a production environment. This configuration is purely for “kicking the tires” or creating a Proof of Concept (PoC) of vSphere with Tanzu. We hope you enjoy it!

Prerequisites

Installation/Configuration

This document assumes you know how to install and configure ESXi and VCSA, enable DRS and HA and configure networking and shared storage. If you are not comfortable with that, then we highly encourage you to take a lab at VMware’s Hands-On Labs. It’s free! Go to Networking

To make networking as easy as possible we recommend the following setup for your PoC environment.

You will need two separate, routable subnets configured. One subnet will be for Management Networking. This is where vCenter, ESXi, the Supervisor Cluster and the Load Balancer will live. The other subnet will be used for Workload Networking. This is where your virtual IP’s and TKG clusters will live.

Note: in this example the management IP of the HAProxy appliance is on the same network as the management network of vCenter and ESXi. This is not a best practice and is only done to simplify the installation and configuration of vSphere with  Tanzu in a PoC mode. In production the HAProxy management IP needs only to be routable to the Supervisor Cluster management IP's.

As you will see, these two subnets are going to be configured on separate portgroups. If these subnets are on separate VLANs then you will need to configure the portgroups accordingly.

Note: The Management and Workload Networks cannot be on the same subnet. They also require L2 isolation. We highly recommend using VLANs to isolate the Management and Workload Networks.

“Simplified” vSphere with Tanzu Network Topology

 Diagram

Description automatically generated

Note: In the documentation you will see mention of a “frontend network” configuration. This guide purposely did not include that configuration in the goal of making this as simple as possible. The Frontend configuration would be used in a production environment to isolate the nodes of your clusters from the network used by developers to access the cluster.

Subnets

The size of each subnet is dependent on your configuration needs. If you are just installing this to “try it out” and have limited subnet resources, the Management network could be very small (see below), and your Workload Network could be as small as a /28. That would give you 14 addresses.

Let’s look at the bare minimum requirements. First, we will start with subnet masks. This will give you an idea of how many IP addresses to request for your evaluation.

Subnet Mask

/28

/27

/26

/25

/24

IP Addresses

14

30

62

162

254

Workload Management Network

The Supervisor Cluster and Load Balancer are “dual homed”. They have a virtual NIC attached to both the Workload Management Network and the Workload Network.

  • This is to allow the Supervisor Cluster to program the load balancer
  • The VM's IP address on this network should be static since the Supervisor Cluster will not be able to program the load balancer if the load balancer's control plane IP address changes.
  • This is also the network to which the VM's default gateway should belong.
  • Finally, other system activity, such as DNS queries, will occur via this network

Note: While the Workload Management network is not required to be on the same network as the ESXi and vCenter, for the purposes of this “PoC” configuration they will be the same. Going forward in the guide the term “Management Network” will apply to the same network.

For PoC purposes the “Management Network” can be on a VSS or a portgroup on a VDS.

Components

Management Network IP Address
Minimal
Requirements

Supervisor
Cluster

5 IPs

ESXi Hosts

1 per host

VCSA

1 IP

Load Balancer

1 IP

Workload Network

The Workload Network has the following characteristics:

  • This network is used by the load balancer to access the Kubernetes services on the Supervisor and Guest clusters.
  • When the HAProxy VM is deployed with only two NICs, the Workload network must also provide the logical networks used to access the load balanced services.

Components

Workload Network IP Address
Minimal
Requirements

Supervisor
Cluster

3 IPs

TKG Cluster
Controller

1 IP per Controller

TKG Cluster
Worker

1 IP per Worker

Load Balancer

1 IP per Kubernetes LB Service

Virtual IPs

TBD

 

The main takeaway here is that there are two ranges of IP addresses in use in the Workload Network subnet.

  • The Cluster Node Range
    • The range for Supervisor and Guest Cluster nodes on the workload network.
    • In the UI during the deployment of the Load Balancer OVA (in this case HAProxy) this is referred to as the Load Balancer IP Ranges

Graphical user interface, text, application

Description automatically generated

  • The Virtual IP Range
    • These are the IP addresses offered up by the Load Balancer that will route to a TKG cluster or application you set up. Developers running Kubectl will connect to their vSphere Namespace and TKG clusters using one of these IP Addresses. These IP Addresses will be provisioned from this range when a developer creates a Kubernetes Service of “Type: Load Balancer”’
    • In the UI during the deployment of the Workload Network this is referred to as “IP Address Ranges for Virtual Servers”

 

Graphical user interface, text, application

Description automatically generated

 

Note: If you have a single TKG cluster with one Control Plane VM and three Worker VMs then you are now at ten IP’s leaving you four VIPs. That’s the absolute bare minimum. We would recommend at least a /27 (30 IP addresses) or a /26 (62 IP Addresses).

 If you are looking to set this up for developers to try, then you probably want a /25 (126 IP Addresses) or a /24 (255 IP Addresses). This will allow them to create several TKG clusters for their testing and validation.

vSphere Distributed Switch Setup

First, you will need a vSphere Distributed Switch (VDS) configured on all the hosts in your cluster. Please see the documentation for how to set that up. This should be a version 7 VDS. (The default)

You will need IP addresses in two separate, routable subnets. The first subnet will be the one that we’ll call “Management”. This is where your VCSA, ESXi hosts, Control Plane VMs and Load Balancer (e.g. HAProxy). On the ESXi hosts this subnet will connect via vmnic0.

This network can live either on a vSphere Standard Switch or a vSphere Distributed Switch portgroup. VM Network will be used as the Management Network.

Next, you will need at least one VDS Portgroup set up. If you have vSphere Standard Switching (VSS) set up already you can use that as your “Management Network”. If you are doing a fresh installation, then this is typically called “VM Network”. This is where you’ll be deploying your load balancer VM to so it should have direct connectivity to ESXi and vCenter.

If you prefer, you can create a separate, new VDS Portgroup and call that “Management”. If you do that and you are using VLANs then ensure that both the VSS and VDS Management portgroup are on the same VLAN and you will have to have a vmkernel adaptor configured with “Management” services configured to ensure proper communication of all components.

Creating the VDS Switch and add all hosts PowerCLI Example

$workloadhosts = get-cluster $Cluster | get-vmhost
#Create the VDS Switch. MTU of 9000 is not necessary.
New-VDSwitch -Name "Dswitch" -MTU 1500 -NumUplinkPorts 1 -location vSAN-DC
Get-VDSwitch "Dswitch" | Add-VDSwitchVMHost -VMHost $workloadhosts
Get-VDSwitch "Dswitch" | Add-VDSwitchPhysicalNetworkAdapter -VMHostNetworkAdapter ($workloadhosts | Get-VMHostNetworkAdapter -Name vmnic2) -Confirm:$false
New-VDPortgroup -Name "Workload Network" -VDSwitch "Dswitch"

Next, you will create a “Workload Network” portgroup. If you are using VLANs then configure this portgroup accordingly.

Graphical user interface, application

Description automatically generated

 

The Workload Network subnet will be “carved” into two IP ranges, one each for the Supervisor Cluster and TKG Cluster systems and one for Load Balancer virtual IPs (VIPs). On the ESXi hosts this subnet will live on vmnic1. This network is required to be on a vSphere Distributed Switch. (Version 7, the default value)

Here’s an example of how the vmnic physical adapters are configured.

Graphical user interface, application

Description automatically generated

Installation

It is recommended that a minimum of three ESXi hosts be used for this configuration. They can be physical hosts or virtualized/nested hosts. As part of the installation we will assume that the hosts have two NIC cards. (vmnic0, vmnic1)

The example used in my lab has three nic cards and vmnic1 is not used.

If you are comfortable setting up vSphere in a nested environment, then you can use that for your proof of concept. It is highly recommended that you reference Nested Virtualization content on William Lam’s website. From there you can subscribe to his Content Library where he has pre-built ESXi virtual machine OVA’s available for installation. The link for his page on nested virtualization is: http://vmwa.re/nestedesxi

Note: Use of nested virtual hosts is not supported in production

ESXi Installation

Physical hosts

Install ESXi on 3 hosts according to the documentation. You will need to use vSphere supported shared storage solution. Typically, this is vSAN, NFS, iSCSI or Fibre Channel. Shared storage is required. Presenting storage volumes directly is not.

Note: vSAN is NOT required for vSphere with Tanzu! Any supported shared storage will work.

ESXi Network Configuration

As guided above, ensure that the ESXi hosts have at least two NICs configured. You can do this with one NIC and using VLANs for isolation but that it outside the scope of this guide.

The vmnic0 card will be the uplink for the “Management Network”. On this network will be the VCSA, ESXi hosts and the Supervisor Control Plane. This network needs access to NTP, DNS and DHCP services.

Note: If you have DHCP enabled on either of the networks then ensure that the ranges used by DHCP are not overlapped by any of the components used by vSphere with Tanzu.

The vmnic1 card will be used as the uplink for the “Workload Network” portgroup on the vDS.

VCSA Installation

Install the VCSA according to the documentation. It should be on the same network as your ESXi hosts. The configuration option to choose for this installation is: VCSA Size: Small

Configuring vCenter

When the VCSA is up and running, log in to administrator@vSphere.local and do the following tasks.

Create a cluster

Enable vSphere HA and DRS on the cluster

  • DRS should be set to fully automated

Add hosts to the cluster

On all hosts, enable vmk0 for all traffic types required. E.g. vMotion, vSAN, etc.

If you are using NFS or iSCSI shared storage, configure this now

If you are using vSAN then configure this now and create a vSAN datastore (ensure vmk0 is enabled for vSAN traffic)

Below is some sample PowerCLI code to help you set vmk0 to the correct settings.

Configure vSwitch and Host Network Adapter PowerCLI Example

# MTU of 9000 is not necessary for vSphere with Tanzu using the vSphere Network Stack
Get-VMHost | Get-VirtualSwitch -Name vSwitch0|Set-VirtualSwitch -mtu 1500 -Confirm:$false
Get-VMHost | Get-VMHostNetworkAdapter -name vmk0|Set-VMHostNetworkAdapter -vSANTrafficEnabled $true -VMotionEnabled $true -Confirm:$false

VDS Configuration

Create a vDS called "Dswitch" (default name) and distributed portgroup called "Workload Network "

Uplink the vDS to vmnic1 on each ESXi host.

Below is an example of some PowerCLI code that can help you automate the proper configuration of the Workload Network for this evaluation. 

All code in this document is given as an example only.

Creating Workload Network VDS Switch and Portgroup PowerCLI Example

# MTU of 9000 is not necessary for vSphere with Tanzu using the vSphere Network Stack
$workloadhosts = get-cluster $Cluster | get-VMHost
New-VDSwitch -Name "Dswitch" -MTU 1500 -NumUplinkPorts 1 -location vSAN-DC
Get-VDSwitch "Dswitch" | Add-VDSwitchVMHost -VMHost $workloadhosts
Get-VDSwitch "Dswitch" | Add-VDSwitchPhysicalNetworkAdapter -VMHostNetworkAdapter ($workloadhosts | Get-VMHostNetworkAdapter -Name vmnic1) -Confirm:$false

Storage Configuration

In this section we are going to create a tagging-based storage profile. The datastore you use needs to be seen by all ESXi hosts in the cluster. When adding the tag, you will also need to create a new tag category. These storage policies will be used in the Supervisor Cluster and namespaces.

 

The policies represent datastores available in the vSphere environment. They control the storage placement of such objects as control plane VMs, pod ephemeral disks, container images, and persistent storage volumes. If you use VMware Tanzu™ Kubernetes Grid™ Service, the storage policies also dictate how the Tanzu Kubernetes cluster nodes are deployed. Let’s get started. The following includes PowerCLI to automate these steps and then a UI version.

Storage Policies PowerCLI Example

# Set up tags for vSphere with Tanzu
$StoragePolicyName = "kubernetes-demo-storage"
$StoragePolicyTagCategory = "kubernetes-demo-tag-category"
$StoragePolicyTagName = "kubernetes-gold-storage-tag"
New-TagCategory -Name $StoragePolicyTagCategory -Cardinality single -EntityType Datastore
New-Tag -Name $StoragePolicyTagName -Category $StoragePolicyTagCategory
Get-Datastore -Name $datastore | New-TagAssignment -Tag $StoragePolicyTagName
New-SpbmStoragePolicy -Name $StoragePolicyName -AnyOfRuleSets (New-SpbmRuleSet -Name "wcp-ruleset" -AllOfRules (New-SpbmRule -AnyOfTags (Get-Tag $StoragePolicyTagName)))

Storage Policies vCenter UI Example

Right-click on the datastore you want to use and select Tags and Custom Attributes and Assign Tags. If you want to use more than one datastore then at the end you can just assign the tag we are about to create to that datastore.

Graphical user interface, text, application, email

Description automatically generated

Click on Add Tag and fill the Tag Name as 'kubernetes-demo-storage'

Graphical user interface, text, application, chat or text message

Description automatically generated

Click create category, enter the Category Name: 'kubernetes-demo-tag-category' then click Create.

Graphical user interface, application

Description automatically generated

 

In the Create Tag box you will see the new tag. Select the category you just created. Click Create.

Graphical user interface, text, application

Description automatically generated

Select this newly created Tag and click Assign.

Graphical user interface, application

Description automatically generated

Now we need to create a tag-based storage policy.

Menu -> Profiles and Policies

 

Graphical user interface, text, application

Description automatically generated

 

VM Storage Policies -> Create VM Storage Policy

Graphical user interface, application, website

Description automatically generated

Name: ‘kubernetes-gold-storage-policy’ then click Next

Graphical user interface, application

Description automatically generated

Select "Enable tag based placement rules " then click Next

Graphical user interface, text, application

Description automatically generated

Tag-based placement -> For Tag Category select kubernetes-demo-tag-category

Graphical user interface, text, application

Description automatically generated

Click browse and select 'kubernetes-demo-storage' then click OK then Next

Graphical user interface, application

Description automatically generated

Under Storage Compatibility you should see the datastore selected in above steps. Click Next

Graphical user interface, text, application

Description automatically generated

Click Finish

Graphical user interface, text, application

Description automatically generated

Storage policies visible to a vSphere Namespace determine which datastores the namespace can access and use for persistent volumes. The storage policies appear as matching Kubernetes storage classes in the namespace. They are also propagated to the Tanzu Kubernetes cluster on this namespace.

Add a DevOps user

If your vCenter is joined to an LDAP or Active Directory you can substitute the “devops” user with a user from that identity store. For the purposes of the demo we are assuming you do not have that available. Instead, we will create a user called “devops” in the vSphere.local identity store. This user will be the one running the kubectl commands later in the document.

To create the devops user do the following:

  1. Menu-> Administration -> Users and Groups
  2. Select Domain -> vSphere.local -> Add User
  3. Add the user "devops" password VMware1! - confirm password -> Add

Graphical user interface

Description automatically generated

Create Content Library

In this step we will add a subscribed Content Library. This Content Library contains the latest TKG cluster images that will be deployed to create TKG clusters. 

Create Content Library PowerCLI Example

#Set up the content library needed by vSphere with Tanzu
New-ContentLibrary -Datastore $datastore -name "tkg-cl" -AutomaticSync -SubscriptionUrl "http://wp-content.vmware.com/v2/latest/lib.json" -Confirm:$false

Create Content Library vCenter UI Example

Do the following: Menu -> Content Libraries

Click “Create”

Graphical user interface, text, application

Description automatically generated

Enter a name and location

Enter the name: tkg-cl

Choose the vCenter Server

Click Next

Graphical user interface, application

Description automatically generated

Configure the Content Library

Select “Subscribed Content Library”

Enter the following URL:

http://wp-content.vmware.com/v2/latest/lib.json
 

Graphical user interface, text, application, email

Description automatically generated

 

Note: This will pull down the latest TKG content directly from VMware. All content is digitally signed and regularly updated.

If you are running systems that are not connected to the Internet there are steps documented in the vSphere documentation on how to get the TKG content.

Click Next

Add Storage

Choose the shared storage option containing your storage policy you configured earlier

Click Next

Graphical user interface, text, application

Description automatically generated

Click Finish

Graphical user interface, text, application, email

Description automatically generated

HAProxy Installation

You are now ready to deploy the HAProxy Load Balancer. Let’s decide on our network configuration and then collect the information we are going to need to accomplish this task.

First, we will need an IP address and DNS address on the Management Network. This must be a static IP.

Management Network Load Balancer IP

10.174.71.50

Management Network Gateway IP

10.174.71.253

 

Note: Deploying this appliance is the equivalent of deploying a piece of L2, networking infrastructure.

For example: The IP range selected for the virtual servers will be reserved by the load balancer appliance. This means if the VIP range is 10.174.72.0/24, and there happens to be a gateway on 10.174.72.253, anyone or anything trying to access a host on 10.174.72.0/24 is going to have a bad time. The appliance will argue it owns 10.174.72.253, any routes that require the gateway 10.174.72.253 failing in the process. Please plan carefully when you decide how you want to configure the Workload Network.

/24 Example

For the example below we are going to use a full /24 subnet for the Workload Network.

Workload Network Load Balancer IP

10.174.72.50

Workload Network Gateway IP

10.174.72.253

Cluster Node Range = 10.174.72.0/25

10.174.72.1-.126

Virtual IP Range        = 10.174.72.208/28

10.174.72.208-.223

We will fine tune down to the IP so that the Load Balancer doesn’t attempt to own the Gateway IP. Based on the values above you will see that we have approximately 124 usable IP addresses set aside for Supervisor Clusters, TKG Clusters, etc.

For the virtual IP range, we have 15 IP addresses set aside.

/25 Example

Workload Network Load Balancer IP

10.174.72.50

Workload Network Gateway IP

10.174.72.1

Cluster Node Range = 10.174.72.0/25

10.174.72.1-.126

Virtual IP Range        = 10.174.72.208/28

10.174.72.209-.223

/26 Example

If a /24 is too much you could go with a smaller subnet and change the values accordingly. For example, let’s say you were given 10.174.72.0/26 which is 62 addresses.

Workload Network Load Balancer IP

10.174.72.31

Workload Network Gateway IP

10.174.72.253

Cluster Node Range = 10.174.72.0/27

10.174.72.3-.30

Virtual IP Range        = 10.174.72.32/27

10.174.72.33-.62

Fine Tuning IPs

To “fine tune” to exclude the gateway or other IP addresses you can put in multiple CIDRs to create “blocks” of IP addresses. For example, if your gateway was 10.174.72.1 and you were given 10.174.72.0/25 as your CIDR range to work with then you would have 126 IP addresses starting at 10.174.72.1 to .126.

To exclude the .1 address and set aside approximately 50 addresses for VIPs you could create some or all the following ranges using CIDR notation.

CIDR

 

 

Usable Ips

10.174.72.8/29

10.174.72.8

10.174.72.15

7

10.174.72.16/28

10.174.72.17

10.174.72.31

15

10.174.72.32/27

10.174.72.33

10.174.72.63

30

Deploy the Load Balancer

For vSphere 7 Update 1 With VDS networking, you need to supply your own load balancer. The first load balancer that is supported is HAProxy. In this section we will deploy it and use some of the values we have talked about above.

Note: You can get a copy of HAProxy from github.com. The location for the HAProxy OVA that has been updated to work with vSphere with Tanzu is here:

https://cdn.haproxy.com/download/haproxy/vsphere/ova/vmware-haproxy-v0.1.8.ova

You can deploy the OVA two ways.

  • Download the OVA using your browser to your desktop and deploy from the vCenter UI
  • Import the OVA directly to a Content Library and deploy from the Content Library
    • See the documentation for this method

Deploy HAProxy PowerCLI Example

If you wish to view the full script using in this document and automate the deployment of the HAProxy OVA and setup of the Content Library, tags and VDS, please check out the PowerCLI script available at the vSphere Tech Marketing Github page here:

https://github.com/vsphere-tmm/Deploy-HAProxy-LB

To deploy from the vCenter UI, right-Click on your cluster and select Deploy OVF Template.

Graphical user interface, application

Description automatically generated

Next, select Local File and click on Upload Files. Find the HAProxy OVA and click Next

Graphical user interface, text, application, email

Description automatically generated

Next, give the VM a name and select where you are going to deploy it to in the folder hierarchy. Click Next.

Graphical user interface, application

Description automatically generated

Select the compute resource you are going to deploy to and click Next.

Graphical user interface

Description automatically generated

 

Review the details and click Next

Graphical user interface, application

Description automatically generated

Accept the License Agreements and click Next

In the next screen you are asked to select Default or Frontend Network. For the purposes of the evaluation we will select Default. You can read about the different options on this screen.

Graphical user interface, text, application, email

Description automatically generated

Select the storage you will be using for the VM

Graphical user interface

Description automatically generated

The next screen is where we select the networks used by the Load Balancer. When using HAProxy in the Default configuration the 3rd option of “Frontend” is displayed but not used when configured. Leave it selected to whatever default network is in the dropdown.

For Management, if you are still using “VM Network” for that network then select that.

For Workload Management, select the Workload Management VDS Portgroup we created earlier. Click Next.

Graphical user interface, text, application, email

Description automatically generated

Any selection in Frontend will be ignored when using the Default Configuration.

Customize HAProxy OVA Template

Appliance Configuration

This is the section that requires you to have done your networking homework!

  • Enter a password for the root account
  • Select whether you wish to permit root login
  • If you are using your own TLS certificate, then 1.3 and 1.4 should include the certificate (ca.crt) from which keys will be generated and the CA certificate private key.
    • If you don’t wish to enter these values, then a self-signed certificate will be generated.

For the purpose of the evaluation leave these values blank.

Graphical user interface, application, Teams

Description automatically generated

Network Configuration

Enter the host name for your load balancer

Enter the DNS Address. If more than one, separate them using commas

Enter the Management IP.

  • This is the static IP address of the appliance on the Management Network.
  • You can’t use a DHCP address here.
  • The value must be in CIDR format. E.g. 10.174.71.51/24. This is done in lieu of using a separate netmask. (e.g. 255.255.255.0)

Enter the Management Gateway IP Address

Enter the Workload IP. E.g. 10.174.72.50

Enter the Workload Gateway IP Address

 

Graphical user interface

Description automatically generated

Load Balancing

Enter the Load Balancer IP Ranges. These are the addresses for the virtual IP Addresses or VIPs used by the load balancer. The load balancer will respond to each of these IP addresses so once you select this range you can’t “give them up” to something else.

  • In the example below I’m using 10.174.72.208/28. This gives me 14 addresses for VIPS. In the examples above this was set to 10.174.72.128/25, giving us 126 VIPs.

Enter the Dataplane API Management Port. This is typically 5556. This will be combined with the Management IP address when we set up vCenter.

Enter a username and password for the Load Balancer Dataplane API and click Next

Graphical user interface, text, application, email

Description automatically generated

Ready to Complete

We are now ready to deploy the Load Balancer. Review the values you set and click Finish.

Power on the Load Balancer VM.

Graphical user interface, application, table

Description automatically generated

Enable Workload Management

With vSphere 7 with Tanzu you get a 60-day evaluation period. In order to enable this, you go to Menu…Workload Management and fill in the contact details so that you can receive communication from VMware. The image below shows all the required fields. Clicking on “I have read and accept the VMware End User License Agreement” will validate the entries. Now click Get Started

Graphical user interface, application

Description automatically generated

You can use the Workload Management functionality during a 60-day evaluation period. However, you must assign the Tanzu Edition license to the Supervisor Cluster before the evaluation period expires.

When the evaluation period of a Supervisor Cluster expires, or the Tanzu Edition license expires, as a vSphere administrator you cannot create new namespaces or update the Kubernetes version of the cluster. As a DevOps engineer, you cannot deploy new Tanzu Kubernetes clusters or change the configuration of the existing ones, such as add a new node and similar. You can still deploy workloads on Tanzu Kubernetes clusters, and all existing workloads continue to run as expected. All Kubernetes workloads that are already deployed continue their normal operation.

Workload Management Setup

After you’ve filled out the license or evaluation screen you are presented with the Workload Management setup screen. From here we will set up the networking support. At this stage we have enabled HA/DRS, set up storage policies, deploy the load balancer and set up the content libraries necessary to continue. That leaves us with the network support setup.

Graphical user interface, text, application, email, website

Description automatically generated

Review the content on the screen. Consider downloading the checklist. It is an Excel spreadsheet and is an excellent item to ensure you’ve covered all the bases. Click on Get Started.
 

vCenter Server and Network

Your vCenter should already be selected. Ensure it is correct.

You will see that you have a choice of networking stacks. Because we haven’t loaded NSX-T it will be greyed out and unavailable.

Click Next.

Graphical user interface, text, application

Description automatically generated

 

Select a Cluster

Select the cluster you’re using

Click Next

Graphical user interface, application, Teams

Description automatically generated

Control Plane Size

Select the size of the resource allocation you need for the Control Plane. For the evaluation, Tiny or Small should be enough.

Click Next.

Graphical user interface, application

Description automatically generated

Storage

Here we will select the storage policy we configured previously

Click Next

Graphical user interface, text, application, email

Description automatically generated

 

Load Balancer

In this section we will use some of the data collected during the deployment of the load balancer.

Enter a DNS-compliant, immutable name. No underscores. Use lower case letters a-z, numbers 0-9, and hyphens E.g. “haproxy-local”

Select the type of Load Balancer: HAProxy

Enter the data plane IP Address. This is the Management IP address AND the port number. In the example here it’s 10.174.72.50:5556.

Enter the username and password used during deployment for the Data plane API user.

Enter the IP Address Ranges for Virtual Server. This is NOT the example used for the VIPs. This is the range of IP addresses that will be used in the Workload Network by TKG clusters.

Finally, enter in the Server CA cert. If you have added a cert during deployment, you would use that. If you have used a self-signed cert then you can retrieve that data from the VM.

  • The easiest method and does not require you to log into the VM is to get the information from the HAProxy VM’s Advanced Settings. See below for a code sample of PowerCLI that will retrieve it.
  • Alternatively, you can get it from the vCenter UIVMRight-clickSettingsVM OptionsAdvancedEdit Configuration guestinfo.dataplaneapi.cacert. However, you will have to convert the string from BASE64. See https://www.base64decode.org for more details.

Graphical user interface, application, Teams

Description automatically generated

HAProxy Certificate Retrieval PowerCLI Code Example

# Change the value of $vc, $vc_user, $vc_password and $VMname to match yours.
$vc = "10.174.71.163"
$vc_user = "administrator@vsphere.local"
$vc_password = "Admin!23"
Connect-VIServer -User $vc_user -Password $vc_password -Server $vc
$VMname = "haproxy-demo"
$AdvancedSettingName = "guestinfo.dataplaneapi.cacert"
$Base64cert = get-vm $VMname |Get-AdvancedSetting -Name $AdvancedSettingName
while ([string]::IsNullOrEmpty($Base64cert.Value)) {
Write-Host "Waiting for CA Cert Generation... This may take a under 5-10 minutes as the VM needs to boot and generate the CA Cert (if you haven't provided one already)."
$Base64cert = get-vm $VMname |Get-AdvancedSetting -Name $AdvancedSettingName
Start-sleep -seconds 2
}
Write-Host "CA Cert Found... Converting from BASE64"
$cert = [Text.Encoding]::Utf8.GetString([Convert]::FromBase64String($Base64cert.Value))
Write-Host $cert

Management Network

Select the network used for the Management Network. In this case I’m selecting “VM Network”

Enter the Starting IP Address.

  • This is the first IP in a range of 5 IPs to assign to Supervisor control plane VMs' management network interfaces.

1 IP is assigned to each of the 3 Supervisor control plane VMs in the cluster

1 IP is used for a Floating IP

 1 is reserved for use during upgrade.

Enter the subnet mask of the Management Network

Enter the Gateway IP address

Enter your DNS server(s)

Optionally, enter your DNS Search Domains

Enter your NTP Server

Click Next

Graphical user interface, application, Teams

Description automatically generated

Workload Network

Here we are going to add your DNS Server and click on Add to start the process of adding the Workload Network. Typically, you can take the default network subnet for “IP Address for Services. Only change this if you are using that subnet elsewhere. This subnet is used for internal communication and it not routed.

  Click Add

Graphical user interface, application

Description automatically generated

 

Adding the Workload Network

Either create a new name or select the default

Select the Workload Network Port Group on the vDS (Dswitch)

Add the gateway for the Workload Network. To follow the worksheet above that would be 10.174.72.253

Enter the subnet mask. For a /24 that is 255.255.255.0

Enter the IP ranges used by resources like TKG clusters on this network. This is the “Cluster Node Range” referred to above. If you selected the whole /24 for your Workload Network when you configured your Load Balancer, then here is where you would be able to isolate out specific addresses by providing a range. To make things simple, let’s put in 10.174.72.100-10.174.72.200.

Click Save

Graphical user interface, text, application

Description automatically generated

Now Click Next and we will move on to TKG Configuration

TKG Configuration

Click on Add

Graphical user interface, text

Description automatically generated

 

Select the TKG Content Library we added previously

Click OK

Click Next

Graphical user interface, application

Description automatically generated

Click Next

Review and Confirm

Click Finish

Graphical user interface, text, application

Description automatically generated

 

During the process of configuring you will see the occasional message become available, updating you on the status of the configuration process. This will take a variable amount of time as several Supervisor Control Plane virtual machines are being provisioned.

 Monitoring Workload Network Configuration

During this process you will see a Namespaces folder be created and the Supervisor Control Plane virtual machines being provisioned into that folder.

 Graphical user interface, application

Description automatically generated

You can monitor the deployment of the VM’s in the Tasks view for the vSphere Cluster. You may see some http errors from time to time. Not to worry, the Supervisor Cluster will keep retrying.

You can monitor the status of the configuration by watching the Tasks and Events pane in the vCenter UI for the vSphere Cluster you enabled Workload Management on.

Graphical user interface

Description automatically generated

If you go to Workload Management…Namespaces you will see this screen until configuration has completed. This can take a while (20+ minutes or more).

A close up of a person

Description automatically generated

While you are waiting, notice that the Supervisor Control Plane VM’s are somewhat unique. You should not modify or change them in any way. They are managed by vCenter.

 Graphical user interface, text, application

Description automatically generated

Create a vSphere Namespace

When the system is ready you will see under Workload ManagementNamespaces the following screen:

Diagram

Description automatically generated

Click on Create Namespace.

Namespace Configuration

Select the cluster

Enter a name for your Namespace

Select the network your Namespace will use

Optionally add a description

Click “Create”

Graphical user interface, text, application

Description automatically generated

 

You will now be presented with the Namespace page for your new Namespace. Click on “Got It”.

Graphical user interface, text, application, email

Description automatically generated

At this point you will configure several Namespace options

Add Permissions

Add Storage

Configure Capacity and Usage

Graphical user interface, application

Description automatically generated

Let’s break each one of these down. First, let’s start with the link to the CLI tools. The CLI tools are the kubectl command that is the key method for developers to interact with Kubernetes. From here you can download a copy of Kubectl that will speak to vSphere. Click on Open and follow the instructions on downloading and installing Kubectl on your client OS.

Note: Depending on your network configuration, you may need a system with a web browser on this network in order to access this web page.

Graphical user interface, text, application

Description automatically generated

Namespace Permissions

Go back to the Namespace and click on Permissions. Here we will grant the previously created “devops” user the Edit permission on the namespace.

Select the Identity Source (vsphere.local)

Enter the username (devops)

Select the Role (Edit)

A picture containing graphical user interface

Description automatically generated

Add Storage to the Namespace

Click on Add Storage

Select the Kubernetes-demo-storage policy created earlier

Click OK

Graphical user interface, table

Description automatically generated

Edit Namespace Resource Limits

Optionally you can edit the resource limits. After all, a Namespace IS a Resource Pool! For the purposes of this document and exercise, let’s hold off on limits for now.

Click on Edit Limits

View the dialog box and optionally adjust the limits used by this namespace.

Click OK

A picture containing table

Description automatically generated

Use Case Examples

You are now ready for your first deployment of a TKG cluster!

Login as devops user

Let’s confirm that you can login. From the system you have installed kubectl, enter the following:

kubectl vsphere login --server=https://10.174.72.209 --vsphere-username devops@vsphere.local --insecure-skip-tls-verify

This is NOT the vCenter Server IP address. This is the IP address referenced in the Namespace.

Enter the password you chose for the devops user.

Password:

You should see the following response:

Logged in successfully.

You have access to the following contexts:

   10.174.72.209

   devops

If the context you wish to use is not in this list, you may need to try

logging in again later or contact your cluster administrator.

To change context, use

kubectl config use-context <workload name>

Now change your kubectl context to the namespace you created

kubectl config use-context devops

Switched to context "devops"

Where to go for more on using Kubernetes

You are now ready to move on to the VMware’s Github where we have shared a set of instructions on how to deploy TKG cluster workloads in your new environment. Please see the link below.

Deploy a workload on the TKC Cluster

https://github.com/vsphere-tmm/vsphere-with-tanzu-quick-start

Share with users/developers

Ultimately you want to have your development team try out your new PoC. Create a namespace for them, give them permissions, set resources and share with them the IP address to download the kubectl binary and the IP address to connect kubectl to the PoC. You can then share with them the GitHub page that we have created, and they can try the example there or they can start uploading their own code to try out.

Next Steps

Now that you have a working Proof of Concept up and running you may want to now consider how you are going to enable vSphere with Tanzu in your existing vSphere installations.

As you’ve discovered in this exercise, your biggest challenge was probably the networking. The big takeaway is to plan, plan and then plan again.

Ensuring your networking configuration is ready to be used by vSphere with Tanzu is key to a successful rollout.

Ensuring you have your subnets, routers, gateways & VLANs all documented before deploying the load balancer and enabling Workload Management is also key.

Because so many networks are set up differently it is imperative that you work with your networking team to make this PoC a success.

In closing

We would like to take this opportunity to thank you for getting this far in this “quick” start guide. If you have feedback, please send it via Twitter to @mikefoley and @mylesagray. We will be updating this document based on your feedback.

 

Filter Tags

  • Modern Applications
  • Intermediate
  • Quick-Start
  • Document
  • vSphere
  • vSphere 7
  • vSphere with Tanzu
  • Kubernetes
  • Deploy