Skip to content

Latest commit

 

History

History
445 lines (361 loc) · 14.5 KB

README.md

File metadata and controls

445 lines (361 loc) · 14.5 KB
Raw

Kubernetes 1.28.3 Swift Package Manager macOS + iOS + Linux CI Status

Table of contents

Overview

SwiftkubeModel is a zero-dependency Swift package for Kubernetes API objects.

  • Model structs for all Kubernetes objects
  • Codable support
  • Hashable resources
  • Closure-based builders for convenient object composition
  • Type-erased wrappers for Kubernetes resources
  • UnstructuredResource type for handling any Kubernetes resource

Compatibility Matrix

<1.18.9 1.18.9 - 1.18.13 1.19.8 1.20.9 1.22.7 1.24.8 1.24.10 1.25.9 1.26.4 1.28.0 1.28.3 1.29.6
0.2.x - ✓ - - - - - - - - - -
0.3.x - - ✓ - - - - - - - - -
0.4.x - - - ✓ - - - - - - - -
0.5.x - - - - ✓ - - - - - - -
0.6.x - - - - - ✓ - - - - - -
0.7.x - - - - - ✓ - - - - - -
0.8.x - - - - - - ✓ - - - - -
0.9.x - - - - - - - ✓ - - - -
0.10.x - - - - - - - - ✓ - - -
0.11.x - - - - - - - - ✓ - - -
0.12.x - - - - - - - - - ✓ - -
0.13.x - - - - - - - - - - ✓ ✓
0.14.x - - - - - - - - - - - ✓
  • ✓ Exact match of API objects in both model and the Kubernetes version.
  • - API objects mismatches either due to the removal of old API or the addition of new API. However, everything the
  • model and Kubernetes have in common will work.

Usage

To use the Kubernetes objects just import SwiftkubeModel:

import SwiftkubeModel

let metadata = meta.v1.ObjectMatadata(name: "swiftkube")
let pod = core.v1.Pod(metadata: metadata)

All the objects are namespaced according to their API group and version, e.g. apps.v1.Deployment or networking.v1beta1.Ingress. Which means, that for example rbac.v1.Role and rbac.v1beta1.Role are completely different objects.

Examples

Any Kubernetes object can be constructed directly using the model structs. Here is an example for a Deployment manifest:

let deployment = apps.v1.Deployment(
    metadata: meta.v1.ObjectMeta(
        name: "nginx"
    ),
    spec: apps.v1.DeploymentSpec(
        replicas: 1,
        selector: meta.v1.LabelSelector(
            matchLabels: ["app": "nginx"]
        ),
        template: core.v1.PodTemplateSpec (
            spec: core.v1.PodSpec(
                containers: [
                    core.v1.Container(
                        image: "nginx",
                        name: "nginx",
                    )
                ]
            )
        )
    )
)

Here is a ConfigMap:

let configMap = core.v1.ConfigMap(
    metadata: meta.v1.ObjectMeta(
        name: "config"
    ),
    data: [
        "env": "dev",
        "log_leve": "debug"
    ]
)

A more complete example of a Deployment, that defines Probes, ResourceRequirements, Volumes and VolumeMounts would look something like this:

let deployment = apps.v1.Deployment(
    metadata: meta.v1.ObjectMeta(
        name: "opa",
        namespace: "default"
    ),
    spec: apps.v1.DeploymentSpec(
        replicas: 2,
        selector: meta.v1.LabelSelector(
            matchLabels: ["app": "opa"]
        ),
        template: core.v1.PodTemplateSpec (
            spec: core.v1.PodSpec(
                containers: [
                    core.v1.Container(
                        image: "openpolicyagent/opa",
                        name: "opa",
                        readinessProbe: core.v1.Probe(
                            failureThreshold: 1,
                            httpGet: core.v1.HTTPGetAction(
                                path: "/health",
                                port: 8080
                            ),
                            initialDelaySeconds: 10,
                            periodSeconds: 20,
                            successThreshold: 2,
                            timeoutSeconds: 5
                        ),
                        resources: core.v1.ResourceRequirements(
                            limits: [
                                "ram": "512MB"
                            ],
                            requests: [
                                "ram": "128MB",
                                "cpu": "200m",
                            ]
                        ),
                        volumeMounts: [
                            core.v1.VolumeMount(
                                mountPath: "/etc/test",
                                name: "data"
                            )
                        ]
                    )
                ],
                imagePullSecrets: [
                    core.v1.LocalObjectReference(name: "secret-name")
                ],
                volumes: [
                    core.v1.Volume(
                        name: "data",
                        persistentVolumeClaim: core.v1.PersistentVolumeClaimVolumeSource(
                            claimName: "pvc",
                            readOnly: true
                        )
                    )
                ]
            )
        )
    )
)

Builders

From the above example it is clear, that a certain knowledge of all the subtypes and their API groups is required, in order to comose a complete manifest. Furthermore, Swift doesn't allow arbitrary arguments order.

For this purpose SwiftkubeModel provides simple closure-based builder functions for convenience. All these functions reside under the sk namespace.

⚠️ The syntax is not yet finalized and can break many times before v1.0.0 ships. This can also be replaced with Function/Result Builders, which is currently a WIP.

⚠️ SwiftkubeModel currently provides convenience builders only for the most common Kubernetes objects.

The above example would look like this:

let deployment = sk.deployment(name: "opa") {
    $0.metadata = sk.metadata {
        $0.namespace = "default"
    }
    $0.spec = sk.deploymentSpec {
        $0.replicas = 1
        $0.selector = sk.match(labels: ["app": "nginx"])
        $0.template = sk.podTemplate {
            $0.spec = sk.podSpec {
                $0.containers = [
                    sk.container(name: "opa") {
                        $0.image = "openpolicyagent/opa"
                        $0.readinessProbe = sk.probe(action: .httpGet(path: "/health", port: 8080)) {
                            $0.failureThreshold = 1
                            $0.initialDelaySeconds = 10
                            $0.periodSeconds = 20
                            $0.successThreshold = 2
                            $0.failureThreshold = 5
                        }
                        $0.resources = sk.requirements {
                            $0.requests = [
                                "ram": "512MB"
                            ]
                            $0.limits = [
                                "ram": "128MB",
                                "cpu": "200m",
                            ]
                        }
                        $0.volumeMounts = [
                            sk.volumeMount(name: "data", mountPath: "/etc/test")
                        ]
                    }
                ]
                $0.imagePullSecrets = [
                    sk.localObjectReference(name: "secret-name")
                ]
                $0.volumes = [
                    sk.volume(name: "data", from: .persistentVolumeClaim(claimName: "pvc", readOnly: true))
                ]
            }
        }
    }
}

Extensions

In addition to closure-based builders, SwiftkubeModel extends the Model objects with some convenience functions, inspired by cdk8s

core.v1.ConfigMap

  • Populating a ConfigMap
let configMap: core.v1.ConfigMap = sk.configMap(name: "test")

// populate the config map
configMap.add(data: "stuff", forKey: "foo")
configMap.add(binaryData: <binary>, forKey: "foo")
configMap.add(file: URL(fileURLWithPath: "/some/path"), forKey: "foo")
configMap.add(binaryFile: URL(fileURLWithPath: "/some/path"), forKey: "foo")

core.v1.Container

  • Mount a volume in a container
let container: core.v1.Container = ...
let volume: core.v1.Volume = ...

// mount a volume in a container
container.mount(volume: volume, on: "/data")
container.mount(volume: "dataVolume", on: "/data")

core.v1.Namespace

  • Finalizers
let namespace: core.v1.Namespace = ...

// add/remove finalizers
namespace.add(finalizer: "foo")
namespace.remove(finalizer: "foo")

core.v1.Secret

  • Populating a Secret: the values are Base64-encoded automatically
let secret: core.v1.Secret = sk.secret(name: "test")

// populate the secret
configMap.add(data: "stuff", forKey: "foo")
configMap.add(file: URL(fileURLWithPath: "/some/path"), forKey: "foo")

core.v1.Service

  • Server ports on a service
let service: core.v1.Service = ...

// add a service port entry
service.serve(port: 8080, targetPort: 80)

core.v1.ServiceAccount

  • Use secrets
let serviceAccount: core.v1.ServiceAccount = ...

// add an object reference for a secret
serviceAccount.use(imagePullSecret: "pullSecret")
serviceAccount.use(secret: "secret", namespace: "ns")

apps.v1.Deployment

  • Exposing a Deployment
let deployment: apps.v1.Deployment = ...

// expose a deployment instance to create a service
let service = deployment.expose(on: 8080, type: .clusterIP)

Type-erasure

Often when working with Kubernetes the concrete type of the resource is not known or not relevant, e.g. when creating resources from a YAML manifest file. Other times the type or kind of the resource must be derived at runtime given its string representation.

SwiftkubeModel provides a type-erased resource implementation UnstructuredResource and its corresponding List-Type UnstructuredResourceList in order to tackle these use-cases.

UnstruturedResource allows objects that do not have registered KubernetesAPIResources to be manipulated generically. This can be used to deal with the API objects from a plug-in or CRDs.

Here are some examples to clarify their purpose:

// Given a JSON string, e.g. at runtime, containing some Kubernetes resource
let json = """
  {
    "apiVersion": "stable.example.com/v1",
    "kind": "CronTab",
    "metadata": {
      "name": "my-new-cron-object",
      "namespace": "default"
    },
     "spec": {
       "cronSpec": "* * * * */5",
       "image": "my-awesome-cron-image"
     }
  }
"""

// We can still decode it without knowing the concrete type
let data = str.data(using: .utf8)!
let resource = try? JSONDecoder().decode(UnstructuredResource.self, from: data)

// When encoding the previous instance, it serializes the underlying resource
let encoded = try? JSONEncoder().encode(resource)

The UnstruturedResource exposes its internal dictionary representation and also provides a dynamic subscript support:

let json = """
  {
    "apiVersion": "stable.example.com/v1",
    "kind": "CronTab",
    "metadata": {
      "name": "my-new-cron-object",
      "namespace": "default"
    },
     "spec": {
       "cronSpec": "* * * * */5",
       "image": "my-awesome-cron-image"
     }
  }
"""

let data = str.data(using: .utf8)!
let cron = try? JSONDecoder().decode(UnstructuredResource.self, from: data)

// Shortcut vars
print(cron.apiVersion)
print(cron.kind)
print(cron.metadata)

// The internal Dictionary<String: Any> representation
print(cron.properties)

// Dynamic member lookup
let spec: [String: Any]? = cron.spec
print(spec?["cronSpec"])

Installation

To use the SwiftkubeModel in a SwiftPM project, add the following line to the dependencies in your Package.swift file:

.package(name: "SwiftkubeModel", url: "https://github.com/swiftkube/model.git", from: "0.14.0")

then include it as a dependency in your target:

import PackageDescription

let package = Package(
    // ...
    dependencies: [
        .package(name: "SwiftkubeModel", url: "https://github.com/swiftkube/model.git", from: "0.14.0")
    ],
    targets: [
        .target(name: "<your-target>", dependencies: [
            .product(name: "SwiftkubeModel", package: "SwiftkubeModel"),
        ])
    ]
)

Then run swift build.

License

Swiftkube project is licensed under version 2.0 of the Apache License. See LICENSE for more details.