title: "VMware Tanzu Community Edition Kubernetes Platform in a Homelab" # Title of the blog post.
date: 2022-01-12 # Date of post creation.
# lastmod: 2022-01-06T09:42:51-06:00 # Date when last modified
description: "Gaining familiarity with VMware Tanzu Community Edition by deploying phpIPAM on Kubernetes in my homelab" # Description used for search engine.
featured: false # Sets if post is a featured post, making appear on the home page side bar.
draft: false # Sets whether to render this page. Draft of true will not be rendered.
toc: true # Controls if a table of contents should be generated for first-level links automatically.
usePageBundles: true
# menu: main
# featureImage: "file.png" # Sets featured image on blog post.
# featureImageAlt: 'Description of image' # Alternative text for featured image.
# featureImageCap: 'This is the featured image.' # Caption (optional).
Back in October, VMware [announced](https://tanzu.vmware.com/content/blog/vmware-tanzu-community-edition-announcement) [Tanzu Community Edition](https://tanzucommunityedition.io/) as way to provide "a full-featured, easy-to-manage Kubernetes platform that’s perfect for users and learners alike." TCE bundles a bunch of open-source components together in a modular, "batteries included but swappable" way:
I've been meaning to brush up on my Kubernetes skills so I thought deploying and using TCE in my self-contained [homelab](/vmware-home-lab-on-intel-nuc-9/) would be a fun and rewarding learning exercise - and it was!
Here's how I did it.
### Planning
TCE supports several different deployment scenarios and targets. It can be configured as separate Management and Workload Clusters or as a single integrated Standalone Cluster, and deployed to cloud providers like AWS and Azure, on-premise vSphere, or even a local Docker environment[^yo_dawg]. I'll be using the standard Management + Workload Cluster setup in my on-prem vSphere, so I start by reviewing the [Prepare to Deploy a Cluster to vSphere](https://tanzucommunityedition.io/docs/latest/vsphere/) documentation to get an idea of what I'll need.
Looking ahead, part of the installation process creates a local [KIND](https://kind.sigs.k8s.io/) cluster for bootstrapping the Management and Workload clusters. I do most of my home computing (and homelab work) by using the [Linux environment available on my Chromebook](/setting-up-linux-on-a-new-lenovo-chromebook-duet-bonus-arm64-complications/). Unfortunately I know from past experience that KIND will not work within this environment so I'll be using a Debian 10 VM to do the deployment.
[^yo_dawg]: Yo dawg, I heard you like containers...
#### Networking
The Kubernetes node VMs will need to be attached to a network with a DHCP server to assign their addresses, and that network will need to be able to talk to vSphere. My router handles DHCP for the range `192.168.1.101-250` so I'll plan on using that.
I'll also need to set aside a few static IPs for this project. These will need to be routable and within the same subnet as the DHCP range, but excluded from that DHCP range.
Moving on to the [Getting Started](https://tanzucommunityedition.io/docs/latest/getting-started/), I'll need to grab some software before I can actually Get Started.
#### Kubernetes control plane image
I need to download a VMware OVA which can be used for deploying my Kubernetes nodes from the VMWare Customer Connect portal [here](https://customerconnect.vmware.com/downloads/get-download?downloadGroup=TCE-090)[^register]. There are a few different options available. I'll get the Photon release with the highest Kubernetes version currently available, `photon-3-kube-v1.21.2+vmware.1-tkg.2-12816990095845873721.ova`.
Once the file is downloaded, I'll log into my vCenter and use the **Deploy OVF Template** action to deploy a new VM using the OVA. I won't bother booting the machine once deployed but will rename it to `k8s-node` to make it easier to identify later on and then convert it to a template.
[^register]: Register [here](https://customerconnect.vmware.com/account-registration) if you don't yet have an account.
#### Docker
I've already got Docker installed on this machine, but if I didn't I would follow the instructions [here](https://docs.docker.com/engine/install/debian/) to get it installed and then follow [these instructions](https://docs.docker.com/engine/install/linux-postinstall/#manage-docker-as-a-non-root-user) to enable management of Docker without root.
I also verify that my install is using `cgroup` version 1 as version 2 is not currently supported:
Next up, I'll install `kubectl` [as described here](https://kubernetes.io/docs/tasks/tools/install-kubectl-linux/) - though the latest version is currently `1.23` and that won't work with the `1.21` control plane node image I downloaded from VMware (`kubectl` needs to be within one minor version of the control plane). Instead I need to find the latest `1.22` release.
I can look at the [releases page on GithHub](https://github.com/kubernetes/kubernetes/releases) to see that the latest release for me is `1.22.5`. With this newfound knowledge I can follow the [Install kubectl binary with curl on Linux](https://kubernetes.io/docs/tasks/tools/install-kubectl-linux/#install-kubectl-binary-with-curl-on-linux) instructions to grab that specific version:
It's not strictly a requirement, but having the `kind` executable available will be handy for troubleshooting during the bootstrap process in case anything goes sideways. It can be installed in basically the same was as `kubectl`:
The final bit of required software is the Tanzu CLI, which can be downloaded from the [project on GitHub](https://github.com/vmware-tanzu/community-edition/releases).
I should then be able to access the UI by pointing a web browser at `http://127.0.0.1:8080`... but I'm running this on a VM without a GUI, so I'll need to back up and tell it to bind on `0.0.0.0:8080` so the web installer will be accessible across the network. I can also include `--browser none` so that the installer doesn't bother with trying to launch a browser locally.
*Now* I can point my local browser to my VM and see the UI:
And then I can click the button at the bottom left to save my eyes[^dark_mode] before selecting the option to deploy on vSphere.
I'll plug in the FQDN of my vCenter and provide a username and password to use to connect to it, then hit the **Connect** button. That will prompt me to accept the vCenter's certificate thumbprint, and then I'll be able to select the virtual datacenter that I want to use. Finally, I'll paste in the SSH public key[^gen_key] I'll use for interacting with the cluster.
I click **Next** and move on to the Management Cluster Settings.
This is for a lab environment that's fairly memory-constrained, so I'll pick the single-node *Development* setup with a *small* instance type. I'll name the cluster `tce-mgmt` and stick with the default `kube-vip` control plane endpoint provider. I plug in the control plane endpoint IP that I'll use for connecting to the cluster and select the *small* instance type for the worker node type.
I don't have an NSX Advanced Load Balancer or any Metadata to configure so I'll skip past those steps and move on to configuring the Resources.
Here I pick to place the Tanzu-related resources in a VM folder named `Tanzu`, to store their data on my single host's single datastore, and to deploy to the one-host `physical-cluster` cluster.
This bit is actually pretty easy. For Network Name, I select the vSphere network where the `192.168.1.0/24` network I identified earlier lives, `d-Home-Mgmt`. I leave the service and pod CIDR ranges as default.
I disable the Identity Management option and then pick the `k8s-node` template I had imported to vSphere earlier.
I skip the Tanzu Mission Control piece (since I'm still waiting on access to [TMC Starter](https://tanzu.vmware.com/tmc-starter)) and click the **Review Configuration** button at the bottom of the screen to review my selections.
See the option at the bottom to copy the CLI command? I'll need to use that since clicking the friendly **Deploy** button doesn't seem to work while connected to the web server remotely.
After a moment or two of verifying prerequisites, I'm met with a polite offer to enable Tanzu Kubernetes Grid Service in vSphere:
```
vSphere 7.0 Environment Detected.
You have connected to a vSphere 7.0 environment which does not have vSphere with Tanzu enabled. vSphere with Tanzu includes
an integrated Tanzu Kubernetes Grid Service which turns a vSphere cluster into a platform for running Kubernetes workloads in dedicated
resource pools. Configuring Tanzu Kubernetes Grid Service is done through vSphere HTML5 client.
Tanzu Kubernetes Grid Service is the preferred way to consume Tanzu Kubernetes Grid in vSphere 7.0 environments. Alternatively you may
deploy a non-integrated Tanzu Kubernetes Grid instance on vSphere 7.0.
Note: To skip the prompts and directly deploy a non-integrated Tanzu Kubernetes Grid instance on vSphere 7.0, you can set the 'DEPLOY_TKG_ON_VSPHERE7' configuration variable to 'true'
Do you want to configure vSphere with Tanzu? [y/N]: n
Would you like to deploy a non-integrated Tanzu Kubernetes Grid management cluster on vSphere 7.0? [y/N]: y
```
That's not what I'm after in this case, though, so I'll answer with a `n` and a `y` to confirm that I want the non-integrated TKG deployment.
And now I go get coffee as it'll take 10-15 minutes for the deployment to complete.
Okay, I'm back - and so is my shell prompt! The deployment completed successfully:
```
Waiting for additional components to be up and running...
Waiting for packages to be up and running...
Context set for management cluster tce-mgmt as 'tce-mgmt-admin@tce-mgmt'.
Management cluster created!
You can now create your first workload cluster by running the following:
tanzu cluster create [name] -f [file]
Some addons might be getting installed! Check their status by running the following:
kubectl get apps -A
```
I can run that last command to go ahead and verify that the addon installation has completed:
Excellent! Things are looking good so I can move on to create the cluster which will actually run my workloads.
[^dark_mode]: Enabling dark mode is probably the most important part of this process.
[^gen_key]: If I didn't already have a key pair to use I would generate one with `ssh-keygen -t rsa -b 4096 -C "email@example.com"` and add it to my client with `ssh-add ~/.ssh/id_rsa`.
#### Workload cluster
I won't use the UI for this but will instead take a copy of my `tce-mgmt.yaml` file and adapt it to suit the workload needs (as described [here](https://tanzucommunityedition.io/docs/latest/workload-clusters/)).
Warning: Pinniped configuration not found. Skipping pinniped configuration in workload cluster. Please refer to the documentation to check if you can configure pinniped on workload cluster manually
I can also go into vCenter and take a look at the VMs which constitute the two clusters:
I've highlighted the two Control Plane nodes. They got their IP addresses assigned by DHCP, but [VMware says](https://tanzucommunityedition.io/docs/latest/verify-deployment/#configure-dhcp-reservations-for-the-control-plane-nodes-vsphere-only) that I need to create reservations for them to make sure they don't change. So I'll do just that.
Excellent, I've got a Tanzu management cluster and a Tanzu workload cluster. What now?
[^i_wont]: I'm not going to, but I totally could.
### Working with Tanzu
If I run `kubectl get nodes` right now, I'll only get information about the management cluster:
To be able to deploy stuff to the workload cluster, I need to tell `kubectl` how to talk to it. And to do that, I'll first need to use `tanzu` to capture the cluster's kubeconfig:
Before I move on to deploying actually *useful* workloads, I'll start with deploying a quick demo application as described in William Lam's post on [Interesting Kubernetes application demos](https://williamlam.com/2020/06/interesting-kubernetes-application-demos.html). `yelb` is a web app which consists of a UI front end, application server, database server, and Redis caching service so it's a great little demo to make sure Kubernetes is working correctly.
I can check out the sample deployment that William put together [here](https://github.com/lamw/vmware-k8s-app-demo/blob/master/yelb.yaml), and then deploy it with:
This demo is using a `NodePort` type service to expose the front end, which means it will be accessible on port `30001` on the node it's running on. I can find that IP by:
So I can point my browser at `http://192.168.1.145:30001` and see the demo:
After marveling at my own magnificence[^magnificence] for a few minutes, I'm ready to move on to something more interesting - but first, I'll just delete the `yelb` namespace to clean up the work I just did:
Now let's move on and try to deploy `yelb` behind a `LoadBalancer` service so it will get its own IP. William has a [deployment spec](https://github.com/lamw/vmware-k8s-app-demo/blob/master/yelb-lb.yaml) for that too.
Wait a minute. That external IP is *still*`<pending>`. What gives? Oh yeah I need to actually deploy and configure a load balancer before I can balance anything. That's up next.
[^magnificence]: Mr. Anderson.
#### Deploying `kube-vip` as a load balancer
Fortunately, William Lam [wrote up some tips](https://williamlam.com/2021/10/quick-tip-install-kube-vip-as-service-load-balancer-with-tanzu-community-edition-tce.html) for handling that too. It's [based on work by Scott Rosenberg](https://github.com/vrabbi/tkgm-customizations). The quick-and-dirty steps needed to make this work are:
#### Persistent Volume Claims, Storage Classes, and Storage Policies
At some point, I'm going to want to make sure that data from my Tanzu workloads stick around persistently - and for that, I'll need to [define some storage stuff](https://tanzucommunityedition.io/docs/latest/vsphere-cns/).
First up, I'll add a new tag called `tkg-storage-local` to the `nuchost-local` vSphere datastore that I want to use for storing Tanzu volumes:
Then I create a new vSphere Storage Policy called `tkg-storage-policy` which states that data covered by the policy should be placed on the datastore(s) tagged with `tkg-storage-local`:
So that's the vSphere side of things sorted; now to map that back to the Kubernetes side. For that, I'll need to define a Storage Class tied to the vSphere Storage profile so I drop these details into a new file called `vsphere-sc.yaml`:
I can test that I can create a Persistent Volume Claim against the new `vsphere` Storage Class by putting this in a new file called `vsphere-pvc.yaml`:
Demos are all well and good, but how about a real-world deployment to tie it all together? I've been using a [phpIPAM instance for assigning static IP addresses for my vRealize Automation deployments](/integrating-phpipam-with-vrealize-automation-8/), but have *only* been using it to monitor IP usage within the network ranges to which vRA will provision machines. I recently decided that I'd like to expand phpIPAM's scope so it can keep an eye on *all* the network ranges within the environment. That's not a big ask in [my little self-contained homelab](/vmware-home-lab-on-intel-nuc-9/), but having a single system scanning all the ranges of a large production network probably wouldn't scale too well.
Fortunately the phpIPAM project provides a [remote scanning agent](https://github.com/phpipam/phpipam-agent) which can be used for keeping an eye on networks and reporting back to the main phpIPAM server. With this, I could deploy an agent to each region (or multiple agents to a region!) and divide up the network into chunks that each agent would be responsible for scanning. But that's a pretty lightweight task for a single server to manage, and who wants to deal with configuring multiple instances of the same thing? Not this guy.
So I set to work exploring some containerization options, and I found [phpipam-docker](https://github.com/phpipam-docker/phpipam-docker). That would easily replicate my existing setup in a trio of containers (one for the web front-end, one for the database back-end, and one with `cron` jobs to run scans at regular intervals)... but doesn't provide a remote scan capability. I also found a [dockerized phpipam-agent](https://github.com/pierrecdn/phpipam-agent), but this one didn't quite meet my needs. It did provide me a base to work off of though so a few days of [tinkering](https://github.com/jbowdre/phpipam-agent-docker) resulted in me publishing my first [Docker image](https://github.com/jbowdre/phpipam-agent-docker/pkgs/container/phpipam-agent). I've still some work to do before this application stack is fully ready for production but it's at a point where I think it's worth doing a test deploy.
To start, I'll create a new namespace to keep things tidy:
So it will need a `Service` exposing the container's port `3306` so that other pods can connect to the database. For my immediate demo, using `type: ClusterIP` will be sufficient since all the connections will be coming from within the cluster. When I do this for real, it will need to be `type: LoadBalancer` so that the agent running on a different cluster can connect. And it will need a `PersistentVolumeClaim` so it can store the database data at `/var/lib/mysql`. It will also get passed an environment variable to set the initial `root` password on the database instance (which will be used later during the phpIPAM install to create the initial `phpipam` database).
Based on that, I can see that my `phpipam-www` pod will need a container running the `phpipam/phpipam-www:1.5x` image, a `Service` of type `LoadBalancer` to expose the web interface on port `80`, a `PersistentVolumeClaim` mounted to `/phpipam/css/images/logo`, and some environment variables passed in to configure the thing. Note that the `IPAM_DATABASE_PASS` variable defines the password used for the `phpipam` user on the database (not the `root` user referenced earlier), and the `IPAM_DATABASE_WEBHOST=%` variable will define which hosts that `phpipam` database user will be able to connect from; setting it to `%` will make sure that my remote agent can connect to the database even if I don't know where the agent will be running.
Here's how I'd adapt that into a structure that Kubernetes will understand:
No exposed ports, no need for persistence - just a base image and a few variables to tell it how to connect to the database and how often to run the scans:
It's got a few additional variables to make it extra-configurable, but still no need for persistence or network exposure. That `IPAM_AGENT_KEY` variable will need to get populated the appropriate key generated within the new phpIPAM deployment, but we can deal with that later.
I'll hold off on the agent container for now since I'll need to adjust the configuration slightly after getting phpIPAM set up, but I will go ahead and check out my work so far:
I'm all for easy so I'll opt for **Automatic database installation**, which will prompt me for the credentials of an account with rights to create a new database within the MariaDB instance. I'll enter `root` and the password I used for the `MYSQL_ROOT_PASSWORD` variable above:
The scan agent isn't going to do anything until it's assigned to a subnet though, so now I head to **Administration > IP related management > Sections**. phpIPAM comes with a few default sections and ranges and such defined so I'll delete those and create a new one that I'll call `Lab`.
Now I can create a new subnet within the `Lab` section by clicking the **Subnets** menu, selecting the `Lab` section, and clicking **+ Add subnet**.
I'll define the new subnet as `192.168.1.0/24`. Once I enable the option to *Check hosts status*, I'll then be able to specify my new `remote-agent` as the scanner for this subnet.
It shows the scanner associated with the subnet, but no data yet. I'll need to wait a few minutes for the first scan to kick off (at the five-minute interval I defined in the configuration).
I still need to do more work to the containerized phpIPAM stack ready for production, but I'm feeling pretty good for having deployed a functional demo of it at this point! And working on this was a nice excuse to get a bit more familiar with Tanzu Community Edition specifically, Kubernetes in general, and Docker (I learned a ton while assembling the `phpipam-agent` image!). I find I always learn more about a new-to-me technology when I have an actual project to do rather than just going through the motions of a lab exercise. Maybe my notes will be useful to you, too.
