You will need to build or load the weaveworks/fluxy image into the Docker daemon, since the deployment does not attempt to pull the image from a registry. If you're using minikube to try things locally, for example, you can do
eval $(minikube docker-env)
make clean build
which will build the image in minikube's Docker daemon, thus making it available to Kubernetes.
The file fluxy-deployment.yaml contains a Kubernetes deployment configuration
that runs the latest image of Fluxy.
kubectl create -f fluxy-deployment.yaml
To make the pod accessible to fluxctl, you can port forward:
kubectl port-forward $(kubectl get pods | grep fluxy | awk '{print $1}') 3030:3030
This will work with the default settings of fluxctl,
and is especially handy with minikube.
To force Kubernetes to run the latest image, kill the pod:
kubectl get pods | grep fluxy | awk '{ print $1 }' | xargs kubectl delete pod
The automation component mutates a Git repository, which requires an SSH access key.
That private key is stored as a Kubernetes secret named fluxy-repo-key.
kubectl delete secret fluxy-repo-key
kubectl create secret generic fluxy-repo-key --from-file=id-rsa=/path/to/id_rsa