Skip to main content

Containers

This library allows deploying arbitrary containers with Wing.

Installation​

Use npm to install this library:

npm i @winglibs/containers

Bring it​

The containers.Workload resource represents a containerized workload.

bring containers;

new containers.Workload(
  name: "hello",
  image: "paulbouwer/hello-kubernetes:1",
  port: 8080,
  readiness: "/",
  replicas: 4,
  env: {
    "MESSAGE" => message,
  }
);

Forwarding​

The workload.forward() method returns an IForward object with a fromXxx() method for each supported handler type.

For example, this is how you can forward cloud.Api requests:

let work = new containers.Workload(...);
let api = new cloud.Api();
api.get("/my_request", work.forward().fromApi());

You can pass an optional route and method to forward() in order to customize the behavior:

work.forward(route: "/your_request", method: cloud.HttpMethod.PUT);

sim​

When executed in the Wing Simulator, the workload is started within a local Docker container.

tf-aws​

To deploy containerized workloads on AWS, we will need an EKS cluster. Unless other specified, a cluster will be automatically provisioned for each Wing application.

However, it a common practice to reuse a single EKS cluster for multiple applications. To reference an existing cluster, you will need to specify the following platform values:

  • eks.cluster_name: The name of the cluster
  • eks.endpoint: The URL of the Kubernetes API endpoint of the cluster
  • eks.certificate: The certificate authority of this cluster.

This information can be obtained from the AWS Console or through the script eks-values.sh:

$ ./eks-values.sh CLUSTER-NAME > values.yaml
$ wing compile -t tf-aws --values ./values.yaml main.w

To create a new EKS cluster, you can use the tfaws.Cluster resource:

eks.main.w:

bring containers;

new containers.tfaws.Cluster() as "my-wing-cluster";

And provision it using Terraform:

wing compile -t tf-aws eks.main.w
cd target/eks.main.tfaws
terraform init
terraform apply
./eks-values.sh my-wing-cluster > values.yaml

This might take a up to 20 minutes to provision (now you see why we want to share it across apps?). The last command will populate values.yaml with the the cluster information needed to deploy workloads.

To connect to this cluster using kubectl, use:

aws eks update-kubeconfig --name my-wing-cluster

Then:

$ kubectl get all
NAME                 TYPE        CLUSTER-IP   EXTERNAL-IP   PORT(S)   AGE
service/kubernetes   ClusterIP   172.20.0.1   <none>        443/TCP   36m

Roadmap​

See Captain's Log in the Wing Slack.

  • EKS as a singleton
  • Container logs to Wing logs
  • Add support for local Dockerfiles (currently only images from Docker Hub are supported), this includes publishing into an ECR.
  • Invalidation of local docker image (both local and in registry). Check what cdk-assets is doing.
  • Reference existing EKS repository.
  • Use a cloud.Redis database
  • Implement cloud.Service using containers.
  • Reference workload from another workload (without going through the load balancer) - Microservice example.
  • internalUrl() in the simulator/aws.
  • publicUrl() in simulator/aws.
  • Logging in tf-aws (Disabled logging because aws-logging configmap was not found. configmap "aws-logging" not found).
  • Logging in sim.
  • Publish the library
  • Generate helm charts under target directory
  • Implement start() and stop().
  • Sidecar containers
  • Domains
  • How can we vend ./eks-value.sh as part of this library?
  • SSL
  • Nodes - what should we do there? Use Fargate profiles in EKS instead of managed node groups?
  • Open bugs
  • Restore microservice test (fails on GitHub).

License​

Licensed under the MIT License.​