GitHub - DefangLabs/pulumi-defang: Defang Pulumi provider...

The Pulumi Providers for Defang — take your app from Docker Compose to a secure, scalable cloud deployment with Pulumi. Develop once, deploy anywhere: the AWS, GCP, and Azure providers share a near-identical Compose-shaped API, so the same program targets a different cloud by swapping a single import.

The repo also contains the source of our CD program in the cd/ directory, which serves as the driver for the Defang deployments: it receives a Compose file, translates it to a Pulumi program, and runs pulumi up (or destroy etc.) to (de)provision the resources in it.

Most users will want to use these components through the Defang CLI, which connects to your cloud account, bootstraps the CD environment and runs the CD image built by this repo.

If you're a Pulumi user, check out our Pulumi Registry listing for installation instructions, API docs, and example usage of using the providers in your own Pulumi programs.

The examples/ directory in this repo also contains complete working samples for all clouds and languages, including examples/multi-cloud which deploys a single Compose app to all three clouds from a single Pulumi program.

Components

Each provider (defang-aws, defang-gcp, defang-azure) exposes the same Pulumi resource palette:

Project — the recommended entry point. Takes a full services map (Compose-style) and provisions shared infrastructure (VPC, networking, DNS, load balancers, build pipelines) alongside each service.
Service — a single container service. Standalone use is image-only: image must refer to a pre-built image. Build-from-source is a Project responsibility because it needs the shared build pipeline (Artifact Registry + Cloud Build on GCP, ECR + CodeBuild on AWS, ACR on Azure).
Build (AWS only) — an image-build resource used internally by Project.
Postgres / Redis — managed database / cache components. Can be used standalone or as part of a Project.

Managed components (Service, Postgres, Redis) share one implementation between standalone and project-scoped use. Standalone instantiations skip the shared-infra provisioning and therefore don't support features that depend on it (VPC access, build-from-source, external load-balancer wiring).

Resource naming

When creating Pulumi resources, these Pulumi components will only specify the logical names, which is either the service name (e.g. app) or a name to describe the resource's role (e.g. shared-vpc, ecr-public). The physical name of the underlying cloud resource is determined by the Pulumi engine. By default, this will be the logical name followed by a hyphen and 7 random hex characters (e.g. app-abc1234). To control the physical name, configure autonaming rules in the Pulumi configuration files, either globally or per resource type. See the Pulumi autonaming docs for details. See the CD code for examples of how autonaming is used in Defang.

Installation and Configuration

See our Installation and Configuration docs

Development

See the Contributing doc.

Example usage

The examples below use AWS. The same code targets GCP or Azure by swapping the import (defang-aws → defang-gcp or defang-azure) — the services map and the rest of the program are unchanged. Complete working samples for all three clouds and four languages are in the ./examples directory, and examples/multi-cloud deploys the same Compose app to all three from a single program.

import * as pulumi from "@pulumi/pulumi";
import * as defang_aws from "@defang-io/pulumi-defang-aws";

const awsDemo = new defang_aws.Project("aws-demo", {services: {
    app: {
        image: "nginx",
        ports: [{
            target: 80,
            mode: "ingress",
            appProtocol: "http",
        }],
    },
}});
export const endpoints = awsDemo.endpoints;

import pulumi
import pulumi_defang_aws as defang_aws

aws_demo = defang_aws.Project("aws-demo", services={
    "app": {
        "image": "nginx",
        "ports": [{
            "target": 80,
            "mode": "ingress",
            "app_protocol": "http",
        }],
    },
})
pulumi.export("endpoints", aws_demo.endpoints)

package main

import (
	defangaws "github.com/DefangLabs/pulumi-defang/sdk/v2/go/defang-aws"
	"github.com/DefangLabs/pulumi-defang/sdk/v2/go/defang-aws/compose"
	"github.com/pulumi/pulumi/sdk/v3/go/pulumi"
)

func main() {
	pulumi.Run(func(ctx *pulumi.Context) error {
		awsDemo, err := defangaws.NewProject(ctx, "aws-demo", &defangaws.ProjectArgs{
			Services: compose.ServiceConfigMap{
				"app": &compose.ServiceConfigArgs{
					Image: pulumi.String("nginx"),
					Ports: compose.ServicePortConfigArray{
						&compose.ServicePortConfigArgs{
							Target:      pulumi.Int(80),
							Mode:        pulumi.String("ingress"),
							AppProtocol: pulumi.String("http"),
						},
					},
				},
			},
		})
		if err != nil {
			return err
		}
		ctx.Export("endpoints", awsDemo.Endpoints)
		return nil
	})
}

using System.Collections.Generic;
using System.Linq;
using Pulumi;
using DefangAws = DefangLabs.DefangAws;

return await Deployment.RunAsync(() =>
{
    var awsDemo = new DefangAws.Project("aws-demo", new()
    {
        Services =
        {
            { "app", new DefangAws.Compose.Inputs.ServiceConfigArgs
            {
                Image = "nginx",
                Ports = new[]
                {
                    new DefangAws.Compose.Inputs.ServicePortConfigArgs
                    {
                        Target = 80,
                        Mode = "ingress",
                        AppProtocol = "http",
                    },
                },
            } },
        },
    });

    return new Dictionary<string, object?>
    {
        ["endpoints"] = awsDemo.Endpoints,
    };
});

name: defang-aws
runtime: yaml
description: Example using defang-aws to deploy services to AWS

plugins:
  providers:
    - name: defang-aws
      path: ../../bin

resources:
  aws-demo:
    type: defang-aws:index:Project
    properties:
      services:
        app:
          image: nginx
          ports:
            - target: 80
              mode: ingress
              appProtocol: http

outputs:
  endpoints: ${aws-demo.endpoints}

License

Dual-licensed:

The provider source in this repo (everything that builds the pulumi-resource-defang-{aws,gcp,azure} binaries and the CD program in cd/) is licensed under the GNU Affero General Public License v3.0 (AGPL-3.0). If you fork the providers or run a modified version as a network service, AGPL's source-disclosure obligations apply.
The generated SDKs in sdk/ (the typed TypeScript, Python, Go, and .NET client libraries that you import into your Pulumi program) are licensed under Apache License 2.0. Embedding them in your application carries no copyleft obligation — they're tooling, not infrastructure-of-your-app.

Practically: if you just consume these providers from your Pulumi program (the common case), you're using Apache-2.0 code. AGPL only enters the picture if you fork the provider engine itself.

推荐订阅源

Hacker News: Show HN