0-README.md

Using Tilt from within a container.

Tilt is great, but it doesn't currently run on Windows. But can we run it in a container on Docker Desktop for Windows instead?

First build a Docker image for Tilt using the Dockerfile in this Gist.

$ docker build -t garethr/tilt .

And then run tilt up. Note that we mount the Docker socket and our Kubernetes configuration.

docker run --rm --net=host -it -v ~/.kube/config:/root/.kube/config -v /var/run/docker.sock:/var/run/docker.sock -v ${PWD}:/code  garethr/tilt up backend

Notes

This mainly works! Though a few issues arose that are worth looking into when time permits:

• --watch doesn't work across the file mount
• tilt demo crashes
• on Docker Desktop the kubernetes config file will point to localhost. For this to work from within the container you currently have to change the server to host.docker.internal

backend.yaml

 apiVersion: v1
 kind: Service
 metadata:
   name: tilt-service   labels:     app: tilt
 spec:
   type: LoadBalancer
   ports:
   - port: 5678
     targetPort: 5678
   selector:
     app: tilt
 ---
 apiVersion: extensions/v1beta1
 kind: Deployment
 metadata:
   name: tilt-deployment
 spec:
   selector:
     matchLabels:
       app: tilt
   template:
     metadata:
       name: tilt-deployment
       labels:
         app: tilt
     spec:
       containers:
       - name: backend
         imagePullPolicy: Always
         image: garethr/tilt-built-me
         args: ["-text", "hello Til2"]
         ports:         - containerPort: 5678

Dockerfile

FROM lachlanevenson/k8s-kubectl

RUN wget https://github.com/windmilleng/tilt/releases/download/v0.1.0/tilt.0.1.0.linux.x86_64.tar.gz && \
     tar xvzf tilt.0.1.0.linux.x86_64.tar.gz && \
     mv tilt /usr/local/bin/tilt
RUN tilt analytics opt in
RUN mkdir /code WORKDIR /code
ENTRYPOINT ["tilt"]

Dockerfile.backend

FROM hashicorp/http-echo

Tiltfile

def backend():
   img = static_build('Dockerfile.backend', 'garethr/tilt-built-me')
   yaml = read_file('backend.yaml')
   return k8s_service(img, yaml=yaml)