Use Draft with Azure Container Service and Azure Container Registry to build and deploy an application to Kubernetes


You are viewing documentation for the old version of the Azure Container Service. Azure Kubernetes Service (AKS) is being updated to add new deployment options, enhanced management capabilities, and cost benefit to Kubernetes on Azure. Visit the AKS documentation to start working with these preview features.

Draft is a new open-source tool that makes it easy to develop container-based applications and deploy them to Kubernetes clusters without knowing much about Docker and Kubernetes -- or even installing them. Using tools like Draft let you and your teams focus on building the application with Kubernetes, not paying as much attention to infrastructure.

You can use Draft with any Docker image registry and any Kubernetes cluster, including locally. This tutorial shows how to use ACS with Kubernetes and ACR to create a live but secure developer pipeline in Kubernetes using Draft, and how to use Azure DNS to expose that developer pipeline for others to see at a domain.

Create an Azure Container Registry

You can easily create a new Azure Container Registry, but the steps are as follows:

  1. Create a Azure resource group to manage your ACR registry and the Kubernetes cluster in ACS.

    az group create --name draft --location eastus
  2. Create an ACR image registry using az acr create and ensure that the --admin-enabled option is set to true.

    az acr create --resource-group draft --name draftacs --sku Basic

Create an Azure Container Service with Kubernetes

Now you're ready to use az acs create to create an ACS cluster using Kubernetes as the --orchestrator-type value.

az acs create --resource-group draft --name draft-kube-acs --dns-prefix draft-cluster --orchestrator-type kubernetes --generate-ssh-keys


Because Kubernetes is not the default orchestrator type, be sure you use the --orchestrator-type kubernetes switch.

The output when successful looks similar to the following.

waiting for AAD role to propagate.done
  "id": "/subscriptions/<guid>/resourceGroups/draft/providers/Microsoft.Resources/deployments/azurecli14904.93snip09",
  "name": "azurecli1496227204.9323909",
  "properties": {
    "correlationId": "<guid>",
    "debugSetting": null,
    "dependencies": [],
    "mode": "Incremental",
    "outputs": null,
    "parameters": {
      "clientSecret": {
        "type": "SecureString"
    "parametersLink": null,
    "providers": [
        "id": null,
        "namespace": "Microsoft.ContainerService",
        "registrationState": null,
        "resourceTypes": [
            "aliases": null,
            "apiVersions": null,
            "locations": [
            "properties": null,
            "resourceType": "containerServices"
    "provisioningState": "Succeeded",
    "template": null,
    "templateLink": null,
    "timestamp": "2017-05-31T10:46:29.434095+00:00"
  "resourceGroup": "draft"

Now that you have a cluster, you can import the credentials by using the az acs kubernetes get-credentials command. Now you have a local configuration file for your cluster, which is what Helm and Draft need to get their work done.

Install and configure draft

  1. Download draft for your environment at and install into your PATH so that the command can be used.
  2. Download helm for your environment at and install it into your PATH so that the command can be used.
  3. Configure Draft to use your registry and create subdomains for each Helm chart it creates. To configure Draft, you need:

    • your Azure Container Registry name (in this example, draftacsdemo)
    • your registry key, or password, from az acr credential show -n <registry name> --output tsv --query "passwords[0].value".

    Call draft init and the configuration process prompts you for the values above; note that the URL format for the registry URL is the registry name (in this example, draftacsdemo) plus Your username is the registry name on its own. The process looks something like the following the first time you run it.

     $ draft init
     Creating /home/ralph/.draft 
     Creating /home/ralph/.draft/plugins 
     Creating /home/ralph/.draft/packs 
     Creating pack go...
     Creating pack python...
     Creating pack ruby...
     Creating pack javascript...
     Creating pack gradle...
     Creating pack java...
     Creating pack php...
     Creating pack csharp...
     $DRAFT_HOME has been configured at /home/ralph/.draft.
     In order to configure Draft, we need a bit more information...
     1. Enter your Docker registry URL (e.g.,,
     2. Enter your username: draftacsdemo
     3. Enter your password: 
     Draft has been installed into your Kubernetes Cluster.
     Happy Sailing!

Now you're ready to deploy an application.

Build and deploy an application

In the Draft repo are six simple example applications. Clone the repo and let's use the Java example. Change into the examples/java directory, and type draft create to build the application. It should look like the following example.

$ draft create
--> Draft detected the primary language as Java with 91.228814% certainty.
--> Ready to sail

The output includes a Dockerfile and a Helm chart. To build and deploy, you just type draft up. The output is extensive, but should be like the following example.

$ draft up
Draft Up Started: 'handy-labradoodle'
handy-labradoodle: Building Docker Image: SUCCESS ⚓  (35.0232s)
handy-labradoodle: Pushing Docker Image: SUCCESS ⚓  (17.0062s)
handy-labradoodle: Releasing Application: SUCCESS ⚓  (3.8903s)
handy-labradoodle: Build ID: 01BT0ZJ87NWCD7BBPK4Y3BTTPB

Securely view your application

Your container is now running in ACS. To view it, use the draft connect command, which creates a secured connection to the cluster's IP with a specific port for your application so that you can view it locally. If successful, look for the URL to connect to your app on the first line after the SUCCESS indicator.


If you receive a message saying that no pods were ready, wait for a moment and retry, or you can watch the pods become ready with kubectl get pods -w and then retry when they do.

draft connect
Connecting to your app...SUCCESS...Connect to your app on localhost:46143
Starting log streaming...
SLF4J: Failed to load class "org.slf4j.impl.StaticLoggerBinder".
SLF4J: Defaulting to no-operation (NOP) logger implementation
SLF4J: See for further details.
== Spark has ignited ...
>> Listening on

In the preceding example, you could type curl -s http://localhost:46143 to receive the reply, Hello World, I'm Java!. When you CTRL+ or CMD+C (depending on your OS environment), the secure tunnel is torn down and you can continue iterating.

Sharing your application by configuring a deployment domain with Azure DNS

You have already performed the developer iteration loop that Draft creates in the preceding steps. However, you can share your application across the internet by:

  1. Installing an ingress in your ACS cluster (to provide a public IP address at which to display the app)
  2. Delegating your custom domain to Azure DNS and mapping your domain to the IP address ACS assigns to your ingress controller

Use helm to install the ingress controller.

Use helm to search for and install stable/traefik, an ingress controller, to enable inbound requests for your builds.

$ helm search traefik
stable/traefik  1.3.0   A Traefik based Kubernetes ingress controller w...

$ helm install stable/traefik --name ingress

Now set a watch on the ingress controller to capture the external IP value when it is deployed. This IP address will be the one mapped to your deployment domain in the next section.

$ kubectl get svc -w
NAME                          CLUSTER-IP     EXTERNAL-IP     PORT(S)                      AGE
ingress-traefik        80:31046/TCP,443:32556/TCP   1h
kubernetes                 <none>          443/TCP                      7h

In this case, the external IP for the deployment domain is Now you can map your domain to that IP.

Map the ingress IP to a custom subdomain

Draft creates a release for each Helm chart it creates -- each application you are working on. Each one gets a generated name that is used by draft as a subdomain on top of the root deployment domain that you control. (In this example, we use as the deployment domain.) To enable this subdomain behavior, you must create an A record for '*.draft' in your DNS entries for your deployment domain, so that each generated subdomain is routed to the Kubernetes cluster's ingress controller.

Your own domain provider has their own way to assign DNS servers; to delegate your domain nameservers to Azure DNS, you take the following steps:

  1. Create a resource group for your zone.

    az group create --name --location eastus
      "id": "/subscriptions/<guid>/resourceGroups/",
      "location": "eastus",
      "managedBy": null,
      "name": "zones",
      "properties": {
        "provisioningState": "Succeeded"
      "tags": null
  2. Create a DNS zone for your domain. Use the az network dns zone create command to obtain the nameservers to delegate DNS control to Azure DNS for your domain.

    az network dns zone create --resource-group --name
      "etag": "<guid>",
      "id": "/subscriptions/<guid>/resourceGroups/zones/providers/Microsoft.Network/dnszones/",
      "location": "global",
      "maxNumberOfRecordSets": 5000,
      "name": "",
      "nameServers": [
      "numberOfRecordSets": 2,
      "resourceGroup": "",
      "tags": {},
      "type": "Microsoft.Network/dnszones"
  3. Add the DNS servers you are given to the domain provider for your deployment domain, which enables you to use Azure DNS to repoint your domain as you want. The way you do this varies by domain provide; delegate your domain nameservers to Azure DNS contains some of the details that you should know.
  4. Once your domain has been delegated to Azure DNS, create an A record-set entry for your deployment domain mapping to the ingress IP from step 2 of the previous section. azurecli az network dns record-set a add-record --ipv4-address --record-set-name '*.draft' -g -z The output looks something like: json { "arecords": [ { "ipv4Address": "" } ], "etag": "<guid>", "id": "/subscriptions/<guid>/resourceGroups/*", "metadata": null, "name": "*.draft", "resourceGroup": "", "ttl": 3600, "type": "Microsoft.Network/dnszones/A" }
  5. Reinstall draft

    1. Remove draftd from the cluster by typing helm delete --purge draft.
    2. Reinstall draft by using the same draft-init command, but with the --ingress-enabled option: bash draft init --ingress-enabled Respond to the prompts as you did the first time, above. However, you have one more question to respond to, using the complete domain path that you configured with the Azure DNS.
  6. Enter your top-level domain for ingress (e.g.

  7. When you call draft up this time, you will be able to see your application (or curl it) at the URL of the form <appname>.draft.<domain>.<top-level-domain>. In the case of this example, bash curl -s Hello World, I'm Java!

Next steps

Now that you have an ACS Kubernetes cluster, you can investigate using Azure Container Registry to create more and different deployments of this scenario. For example, you can create a draft.basedomain.toplevel domain DNS record-set that controls things off of a deeper subdomain for specific ACS deployments.