Skip to content

EDA on macOS#

Do you want to get full EDA experience on your macOS machine? Well, can't judge you!

Typically, the management and automation platforms of EDA' caliber require a ton of resources to run. But that is not the case with EDA! The microservices architecture and reliance on Kubernetes as a deployment platform make it possible to run EDA on a laptop using open-source tools. And macOS-powered machines (even with M chips1) is not an exception!

Playground repository#

We will need the playground repository on our machine to run EDA installation steps. Pull it, as explained in the Try EDA section.

git clone https://github.com/nokia-eda/playground && \
cd playground

When the playground repo is cloned, let's install the CLI tools that we will need to run EDA installation steps.

make download-tools #(1)!

  1. This will download kind, kubectl, kpt, and yq into a tools folder relative to the current working directory.

    Subsequent steps use these versions of the binaries - you may use your own binaries for your own interactions. If you don't have kubectl in your $PATH, then consider copying the kubectl binary from the tools directory to a location in your $PATH to make use of it in the following steps.

The installer is smart enough to download the tools for the right OS/architecture.

macOS prerequisites#

Before we begin, let's ensure that you run macOS Sonoma v14.6.1 or newer, as the older versions of macOS might have issues with Rosetta emulation. The version can be checked with sw_vers -productVersion command in your terminal.

For Apple products with an M-chip the next step prescribes to check that macOS Rosetta virtualization support enabled. This can be done by running the following command:

softwareupdate --install-rosetta

Docker#

Now it is time to install Docker support on your macOS. There are many options available, the most common ones are:

  1. OrbStack Our pick!
  2. Docker Desktop for Mac
  3. Rancher Desktop
  4. Colima

Below you will find short guides on how to install Docker on your macOS using some of the tools mentioned above. If you already have one installed, you can skip to the next section.

OrbStack is a relatively new software that brings Docker support to macOS. The reason we can recommend it is that it has a great UX, has a VM management support, comes with a lightweight k8s cluster and free for personal use.

It is a native macOS app for both Intel and ARM64-based macs, so the installation is as easy as downloading the dmg file and installing at as usual. OrbStack installer will install the Docker CLI on your system, and will provide the Docker VM to run containers and k8s clusters, and enable the kubernetes cluster on the Kubernetes tab:

When OrbStack is installed, you can check the app settings to ensure that you have the sufficient resources allocated to an internal VM that runs docker daemon:

resources

Colima is an open-source, free and lightweight CLI tool that brings container runtimes to macOS.

It has a variety of installation options, with Homebrew being likely the most popular one:

brew install colima

Colima does not provide the Docker CLI client, so you will need to install one separately if you don't have one installed already. Thankfully, it is as easy as running:

brew install docker

Now, Colima can launch a linux/arm64 VM for us; this VM runs the Docker daemon so that we could run KinD with EDA inside. Here is a command to start the VM with 8 vcpu cores and 16 GB of RAM; this should be enough to run the quickstart demo:

colima start --cpu 8 --memory 16 --profile eda \
--vm-type=vz --vz-rosetta --network-address

We can ensure that the VM is running by checking the status:

colima status -e --profile eda
INFO[0000] colima [profile=eda] is running using macOS Virtualization
INFO[0000] arch: aarch64                                
INFO[0000] runtime: docker                              
INFO[0000] mountType: virtiofs                          
INFO[0000] address: 192.168.107.3                       
INFO[0000] socket: unix:///Users/romandodin/.colima/eda/docker.sock 
INFO[0000] cpu: 8                                       
INFO[0000] mem: 16GiB                                   
INFO[0000] disk: 60GiB

Your docker CLI should also "sense" the docker engine availability and set the context to the Colima VM:

docker context ls
NAME           DESCRIPTION                DOCKER ENDPOINT                                    ERROR
colima-eda *   colima [profile=eda]       unix:///Users/romandodin/.colima/eda/docker.sock
Networking and Load Balancer with Colima

The downside of having a VM that runs the Docker daemon is that additional layer of networking is introduced. Hence, the MetalLB Load Balancer that we install in the kind cluster will use the IP range that is not visible from the macOS host.

You don't need Load Balancer to enjoy EDA, since you can always expose the UI and the necessary services using kubectl expose command, but if you want to have the Load Balancer you will have to setup additional routes[^3].

Kubernetes cluster#

With the container runtime installed and running via one of the tools mentioned above, we need to ensure we have a k8s cluster running on our macOS. You typically have two options:

  1. Use the embedded k8s cluster provided by the tool that adds Docker support on your mac
  2. Setup the kind cluster manually using kind/k3s/etc.

The first option might be the easiest way to get started, since it offers more tight integration with the macOS environment, for example by exposing services and providing a LoadBalancer implementation out of the box.

If you're running OrbStack, you can spin up an embedded, one-node, lightweight cluster by checking the Enable Kubernetes cluster box in the settings: k8s-settings

When OrbStack is done with creating a k8s cluster for you, you will be able to use regular cluster management tools like kubectl/k9s/etc to manage your cluster.

kubectl get nodes #(1)!
  1. kubectl is also installed during the make download-tools step.
NAME       STATUS   ROLES                  AGE   VERSION
orbstack   Ready    control-plane,master   28m   v1.29.3+orb1

Now, your cluster is ready to run its first EDA installation!

Should you choose not to use embedded k8s support in the tool of your choice, you can install a KinD cluster manually.

Run the following command from your playground repository to install a KinD cluster:

make kind

Installing EDA#

Install time! When running docker/k8s on mac we have some complications with the networking, as the VM that runs the k8s cluster is not the same as the one where we run our make targets.

That's why we would have not only set the

  • EXT_DOMAIN_NAME: to the domain name or IP of the k8s cluster node (assuming you're running a single node cluster)

  • EXT_IPV4_ADDR: set to the IP address of the cluster node. You can get it with:

    kubectl get nodes -o jsonpath='{.items[0].status.addresses[0].address}'

but also set a pair of no proxy variables set to the cluster cidr of your cluster. You can get the with:

kubectl get nodes -o jsonpath='{.items[0].spec.podCIDR}'

And once all the variables are known, you can start the installation:

EXT_DOMAIN_NAME=eda-api.k8s.orb.local \
EXT_IPV4_ADDR=198.19.249.2 \
EXT_HTTPS_PORT=443 \
NO_PROXY=192.168.194.0/25 \
no_proxy=192.168.194.0/25 \
make try-eda NO_KIND=yes NO_LB=yes

Potential turbulence

You may experience some hiccups during install, for example

  1. Some application is stuck during install
  2. Simulator nodes not starting up
  3. NPP pods not starting up

This is all due to the fact that the majority of the images are running under Rosetta virtualization (they are not available yet in ARM64 arch). The workaround is to restart the eda-ce deployment when things get stuck.

Connecting to the UI/API#

Depending on the tool you're using to run k8s, your method of connecting to the UI will vary.

In OrbStack, you can access the UI by opening https://eda-api.k8s.orb.local/ in your browser.

macOS 15+

If your browser can not resolve the eda-api.k8s.orb.local domain, you may need to add enable Local Network access policy for the browser of your choice. Check the Apple's documentation and this screenshot showing where the settings are located.

ui

Tearing down#

If something goes wrong during installation, or if you want to reinstall, or maybe you finished playing with EDA, feel free to destroy the cluster following the documentation for the tool you're using to run k8s.

To remove the k8s cluster provided by OrbStack run the following command in the terminal:

orb delete k8s

This will remove the VM that backs up the k8s cluster. To bring back the empty k8s cluster run

orb start k8s

and now you can restart the EDA installation.

  1. These performance cores finally have a purpose! ↩

Comments