Compose on Kubernetes

Compose on Kubernetes allows you to deploy Docker Compose files onto a Kubernetes cluster.

Get started
Developing Compose on Kubernetes

More documentation can be found in the docs/ directory. This includes:

Get started

Compose on Kubernetes comes installed on Docker Desktop and Docker Enterprise.

On Docker Desktop you will need to activate Kubernetes in the settings to use Compose on Kubernetes.

Check that Compose on Kubernetes is installed

You can check that Compose on Kubernetes is installed by checking for the availability of the API using the command:

$ kubectl api-versions | grep compose
compose.docker.com/v1beta1
compose.docker.com/v1beta2

Deploy a stack

To deploy a stack, you can use the Docker CLI:

$ cat docker-compose.yml
version: '3.3'

services:

  db:
    build: db
    image: dockersamples/k8s-wordsmith-db

  words:
    build: words
    image: dockersamples/k8s-wordsmith-api
    deploy:
      replicas: 5

  web:
    build: web
    image: dockersamples/k8s-wordsmith-web
    ports:
     - "33000:80"

$ docker stack deploy --orchestrator=kubernetes -c docker-compose.yml hellokube

Developing Compose on Kubernetes

See the contributing guides for how to contribute code.

Pre-requisites

make
Docker Desktop (Mac or Windows) with engine version 18.09 or later
Enable Buildkit by setting DOCKER_BUILDKIT=1 in your environment
Enable Kubernetes in Docker Desktop settings

For live debugging

Debugger capable of remote debugging with Delve API version 2
- Goland run-configs are pre-configured

Debug quick start

Debug install

To build and install a debug version of Compose on Kubernetes onto Docker Desktop, you can use the following command:

$ make -f debug.Makefile install-debug-images

This command:

Builds the images with debug symbols
Runs the debug installer:
- Installs debug versions of API server and Compose controller in the docker namespace
- Creates two debugging LoadBalancer services (unused in this mode)

You can verify that Compose on Kubernetes is running with kubectl as follows:

$ kubectl get all -n docker
NAME                               READY   STATUS    RESTARTS   AGE
pod/compose-7c4dfcff76-jgwst       1/1     Running   0          59s
pod/compose-api-759f8dbb4b-2z5n2   2/2     Running   0          59s

NAME                                      TYPE           CLUSTER-IP       EXTERNAL-IP   PORT(S)           AGE
service/compose-api                       ClusterIP      10.98.42.151     <none>        443/TCP           59s
service/compose-api-server-remote-debug   LoadBalancer   10.101.198.179   localhost     40001:31693/TCP   59s
service/compose-controller-remote-debug   LoadBalancer   10.101.158.160   localhost     40000:31167/TCP   59s

NAME                          DESIRED   CURRENT   UP-TO-DATE   AVAILABLE   AGE
deployment.apps/compose       1         1         1            1           59s
deployment.apps/compose-api   1         1         1            1           59s

NAME                                     DESIRED   CURRENT   READY   AGE
replicaset.apps/compose-7c4dfcff76       1         1         1       59s
replicaset.apps/compose-api-759f8dbb4b   1         1         1       59s

If you describe one of the deployments, you should see *-debug:latest in the image name.

Live debugging install

To build and install a live debugging version of Compose on Kubernetes onto Docker Desktop, you can use the following command:

$ make -f debug.Makefile install-live-debug-images

This command:

Builds the images with debug symbols
Sets the image entrypoint to run a Delve server
Runs the debug installer
- Installs debug version of API server and Compose controller in the docker namespace
- Creates two debugging LoadBalancer services
  - localhost:40000: Compose controller
  - localhost:40001: API server
The API server and Compose controller only start once a debugger is attached

To attach a debugger you have multiple options:

Use GoLand: configuration can be found in .idea of the repository
- Select the Debug all config, setup breakpoints and start the debugger
Set your Delve compatible debugger to point to use locahost:40000 and localhost:40001
- Using a terminal: dlv connect localhost:40000 then type continue and hit enter

To verify that the components are installed, you can use the following command:

$ kubectl get all -n docker

To verify that the API server has started, ensure that it has started logging:

$ kubectl logs -f -n docker deployment.apps/compose-api compose
API server listening at: [::]:40000
ERROR: logging before flag.Parse: I1207 15:25:13.760739      11 plugins.go:158] Loaded 2 mutating admission controller(s) successfully in the following order: NamespaceLifecycle,MutatingAdmissionWebhook.
ERROR: logging before flag.Parse: I1207 15:25:13.763211      11 plugins.go:161] Loaded 1 validating admission controller(s) successfully in the following order: ValidatingAdmissionWebhook.
ERROR: logging before flag.Parse: W1207 15:25:13.767429      11 client_config.go:552] Neither --kubeconfig nor --master was specified.  Using the inClusterConfig.  This might not work.
ERROR: logging before flag.Parse: W1207 15:25:13.851500      11 genericapiserver.go:319] Skipping API compose.docker.com/storage because it has no resources.
ERROR: logging before flag.Parse: I1207 15:25:13.998154      11 serve.go:116] Serving securely on [::]:9443

To verify that the Compose controller has started, ensure that it is logging:

kubectl logs -f -n docker deployment.apps/compose
API server listening at: [::]:40000
Version:    v0.4.16-dirty
Git commit: b2e3a6b-dirty
OS/Arch:    linux/amd64
Built:      Fri Dec  7 15:18:13 2018
time="2018-12-07T15:25:19Z" level=info msg="Controller ready"

Reinstall default

To reinstall the default Compose on Kubernetes on Docker Desktop, simply restart your Kubernetes cluster. You can do this by deactivating and then reactivating Kubernetes or by restarting Docker Desktop. See the contributing and debugging guides.

Deploying Compose on Kubernetes

Guide for Azure AKS.
Guide for GKE.
Guide for Microk8s.
Guide for Minikube.

bhanditz / compose-on-kubernetes Goto Github PK