Skip to main content
Version: Latest

Working with WebAssembly

Rancher Desktop 1.13.0 added experimental support for running WebAssembly (Wasm) applications. This feature needs to be enabled in Preferences > Container Engine > General.

warning

Note that when using the moby container engine, enabling the Wasm feature switches to a different image store, so previously built or downloaded images will not be available and must be built or downloaded again. The images are not lost; Rancher Desktop will switch back to the old image store when Wasm is disabled again.

Managing containerd Wasm shims​

Running WebAssembly applications on a container runtime requires a specific containerd "shim" for each Wasm runtime/framework used.

Rancher Desktop 1.13 comes bundled with the containerd-spin-shim-v2 shim preinstalled. Future releases are expected to download additional shims automatically when the feature is enabled.

For now additional shims can be installed by the user into the containerd-shims cache directory on the host. The location is

  • Linux: ~/.local/share/rancher-desktop/containerd-shims
  • macOS: ~/Library/Application Support/rancher-desktop/containerd-shims
  • Windows: %LOCALAPPDATA%\rancher-desktop\containerd-shims

Any shim installed there will automatically be copied into the VM and configured for the container engine when Rancher Desktop is started (installing a newer version of the spin shim will override the bundled version).

Developing Wasm Applications with Rancher Desktop​

Developing Wasm applications on your local machine on top of Rancher Desktop typically involves below steps:

  • Create an application in your programming language of choice. You can compile code written in many languages, such as C, C++, Rust, Go, and others, into Wasm modules
  • Compile the code into a Wasm module
  • Package the Wasm module as a OCI container image and push to a container registry
  • Run the Wasm container and/or
  • Deploy the Wasm container into Kubernetes

Creating a Simple App and Compiling It Into a Wasm Module​

You can use the Spin framework from Fermyon to quickly bootstrap a simple Wasm app. Install Spin on your machine following the instructions on the Installing Spin page.

Once you have successfully installed Spin, create a new app via the command spin new.

Select the language you would like to use for development.

$spin new
Pick a template to start your application with:
http-js (HTTP request handler using Javascript)
> http-ts (HTTP request handler using Typescript)

Give a name to your app

$spin new
Pick a template to start your application with: http-ts (HTTP request handler using Typescript)
Enter a name for your new application: rd-spin-hello-world

Provide an optional description and leave the API route path to default

$spin new
Pick a template to start your application with: http-ts (HTTP request handler using Typescript)
Enter a name for your new application: rd-spin-hello-world
Description []: This is a simple typescript app that will be compiled into a Wasm module and run as a Wasm container
HTTP path: /...

Once the command ran successfully, you should see a directory created with the boilerplate code for the Spin app.

Update the index.ts file to return a different message than the default.

import { HandleRequest, HttpRequest, HttpResponse } from "@fermyon/spin-sdk"

export const handleRequest: HandleRequest = async function (request: HttpRequest): Promise<HttpResponse> {
return {
status: 200,
headers: { "content-type": "text/plain" },
body: "Hello from Wasm container!"
}
}

Change to the app directory and run the spin build command to compile the app code into a Wasm module.

$spin build
Building component rd-spin-hello-world with `npm run build`

$ rd-spin-hello-world@1.0.0 build
$ npx webpack --mode=production && mkdir -p target && spin js2wasm -o target/rd-spin-hello-world.wasm dist/spin.js

asset spin.js 4.57 KiB [compared for emit] (name: main)
runtime modules 670 bytes 3 modules
./src/index.ts 2.86 KiB [built] [code generated]
webpack 5.91.0 compiled successfully in 1355 ms

Starting to build Spin compatible module
Spin compatible module built successfully
Finished building all Spin components

Once the build command ran successfully, you should see the rd-spin-hello-world.wasm module created inside the target directory.

Package the Wasm Module as an OCI Container Image and Push to a Container Registry​

Create a Dockerfile with below code to package the Wasm module as a docker image.

FROM scratch
COPY spin.toml /spin.toml
COPY /target/rd-spin-hello-world.wasm /target/rd-spin-hello-world.wasm

Run the command below to package the Wasm module as a container image.

nerdctl build \
--namespace k8s.io \
--platform wasi/wasm \
-t ghcr.io/rancher-sandbox/rd-spin-hello-world:0.1.0 .

Push the image to the container registry

nerdctl push ghcr.io/rancher-sandbox/rd-spin-hello-world:0.1.0

Running the Wasm Container​

Running a Wasm container directly is currently only supported with the moby container engine; there is a bug in nerdctl that prevents it from working with containerd. Ensure you have selected dockerd(moby) as the container engine under Preferences > Container Engine > General to work through the steps in this section.

The following command runs the rd-spin-hello-world sample spin application, built in the previous section, on the moby engine (note the final / on the last line; it is the command to run, and docker run will fail if it is omitted):

docker run \
--detach \
--name spin-demo \
--runtime io.containerd.spin.v2 \
--platform wasi/wasm \
--publish 8080:80 \
ghcr.io/rancher-sandbox/rd-spin-hello-world:0.1.0 \
/

The internal port 80 has been mapped to 8080 and can be tested from the host:

$ curl http://localhost:8080/
Hello from Wasm container!

Running Wasm Apps with Kubernetes​

Running WebAssembly applications on Kubernetes is currently only supported with the containerd runtime; it doesn't work with the cri-dockerd shim used to run Kubernetes on top of moby.

Create a deployment for the sample Wasm container image built in the previous section:

kubectl apply --filename - <<EOF
apiVersion: apps/v1
kind: Deployment
metadata:
name: hello-spin
spec:
replicas: 1
selector:
matchLabels:
app: hello-spin
template:
metadata:
labels:
app: hello-spin
spec:
runtimeClassName: spin
containers:
- name: hello-spin
image: ghcr.io/rancher-sandbox/rd-spin-hello-world:0.1.0
command: ["/"]
EOF

It should print

deployment.apps/hello-spin created

Then create a ClusterIP service and a Traefik ingress contoller:

kubectl apply --filename - <<EOF
apiVersion: v1
kind: Service
metadata:
name: hello-spin
spec:
type: ClusterIP
selector:
app: hello-spin
ports:
- port: 80
---
apiVersion: networking.k8s.io/v1
kind: Ingress
metadata:
name: hello-spin
annotations:
traefik.ingress.kubernetes.io/router.entrypoints: web
spec:
rules:
- host: localhost
http:
paths:
- path: /
pathType: Prefix
backend:
service:
name: hello-spin
port:
number: 80
EOF

Which will print

service/hello-spin created
ingress.networking.k8s.io/hello-spin created

Testing it from the host:

$ curl http://localhost/
Hello from Wasm container!

Ingress IP on Windows​

info

On Windows using localhost for the Traefik ingress will not work.

Instead the ingress IP address should be determined from the Traefik loadbalancer:

C:\>kubectl get service traefik --namespace kube-system --output "jsonpath={.status.loadBalancer.ingress[0].ip}"
192.168.127.2

The sslip.io "magic" DNS service can be used to create a corresponding DNS name for it: 192.168.127.2.sslip.io. This name should be used instead of localhost in the Ingress spec host field.

After deploying the deployment, service, and ingress the app should be available under this domain name:

C:\>curl http://192.168.127.2.sslip.io/hello
Hello world from Spin!