Happy Path Deployment Tool

Visit the Happy Path documentation for more details: https://chanzuckerberg.github.io/happy/

The Happy Path Deployment Tool is an open-source project led by the Chan Zuckerberg Initiative (CZI). It is a platform for deploying and managing containerized applications at scale in adherence to CZI security practices. The tool is designed to be easy to use and operate, and it is available for both on-premises and cloud deployments. Happy builds and deploys your application, and once it is released, helps you support it.

Happy Path is based on these principles:

• Operational simplicity: Happy Path takes the bite out of a complex container orchestration operations and infrastructure management
• Extensibility: Happy Path functionality can be extended through various Terraform hooks, custom Terraform modules and Helm Chart customization
• Reliability: Happy Path is reliable and is production-ready, used by multiple engineering teams at CZI

Security

Please note: If you believe you have found a security issue, please responsibly disclose by contacting us at [email protected]

Repository structure

This project is a monorepo for the components of a Happy ecosystem:

• cli/ - the happy CLI tool
• api/ - the happy API server
• shared/ - components shared between api, cli and the terraform/provider
• hvm/ - the happy version manager
• terraform/ - a collection of TF modules we use to provision long-lived infrastructure and application stacks
• terraform/provider - happy terraform provider
• helm-charts/stack - an experimental helm chart
• examples/ - sample applications that illustrate various concepts that happy supports, such as sidecars, tasks, multi-service deployments, GPU intense workloads, and so on

Features

• Manages short-lived infrastructure (we deploy into your compute)
• Groups services together (we call it a stack for co-deployment), each stack is isolated, and you can have multiple stacks created for the same application.
• Easily promote changes from lower to higher environments
• Supports automated deployments through github workflows
• Has an extensive set of Github workflows
• Supports both AWS ECS and EKS runtimes, and allows for an easy migration between the two
• Abstracts out IaC code with the intent that developers should only focus on the application code
• Supports Linkerd service mesh, mTLS and service-to-service authorization when deployed on EKS with Linkerd installed
• Plays nicely with external-dns, karpenter, cluster-autoscaler
• Integrates with Datadog for dashboarding (assuming you have a datadog agent deployed into your EKS)
• Provides service discovery and autoscaling capabilities
• Supports both amd64 and arm64 architectures
• Supports metrics collection, health monitoring through healthchecks, and synthetics
• Supports rolling updates to minimize downtime

Prerequisites

You will need to have Docker desktop, AWS CLI, and terraform installed to use Happy. You can install them and other useful tools by running

brew tap chanzuckerberg/tap
brew install awscli helm kubectx kubernetes-cli aws-oidc linkerd jq terraform
brew install --cask docker

Docker Desktop needs to be running; and aws cli needs to be configured by running aws configure, with profiles setup. Make sure cat ~/.aws/config does not return an empty string (we assume you already have an AWS account).

In addition to the above, you will need an up and running EKS cluster, that contains a happy environment namespace (it contains a secret called integration-secret).

Integration secret can be set up via happy-env-eks terraform module,

module "happy_env" {
  source = "../../happy-env-eks"
  eks-cluster = {
    cluster_id              = "my-eks-cluster",
    cluster_arn             = "arn:aws:eks:us-west-2:00000000000:cluster/my-eks-cluster",
    cluster_endpoint        = "https://A1B2C3D4.gr7.us-west-2.eks.amazonaws.com",
    cluster_ca              = "...",
    cluster_oidc_issuer_url = "https://oidc.eks.us-west-2.amazonaws.com/id/A1B2C3D4",
    cluster_version         = "1.27",
    worker_iam_role_name    = "my-eks-cluster-eks-node-role-name",
    worker_security_group   = "my-eks-cluster-worker-security-group",
    oidc_provider_arn       = "arn:aws:iam::00000000000:oidc-provider/oidc.eks.us-west-2.amazonaws.com/id/A1B2C3D4",
  }
  okta_teams   = []
  base_zone_id = "ROUTE53_EXTERNAL_ZONE_ID"
  cloud-env = {
    database_subnet_group = "db-subnet-group"
    database_subnets      = ["subnet-xxxxxxxxxxxxxxxxx"...]
    private_subnets       = ["subnet-xxxxxxxxxxxxxxxxx"...]
    public_subnets        = ["subnet-xxxxxxxxxxxxxxxxx"...]
    vpc_cidr_block        = "10.0.0.0/16"
    vpc_id                = "vpc-xxxxxxxxxxxxxxxxx"
  }
  tags = {
    project   = "happy"
    env       = "rdev"
    service   = "happy"
    owned_by  = "happy"
  }
  providers = {
    aws.czi-si = aws.czi-si
  }
}

provider "aws" {
  alias = "czi-si"
}

This module will create a namespace

Another approach is to create the secret explicitly. Create a file called integration-secret.json with the following content:

{
    "kind": "k8s",
    "cloud_env": {
        "database_subnet_group": "db-subnet-group",
        "database_subnets": [
            "subnet-xxxxxxxxxxxxxxxxx",
            "subnet-xxxxxxxxxxxxxxxxx",
            "subnet-xxxxxxxxxxxxxxxxx",
            "subnet-xxxxxxxxxxxxxxxxx"
        ],
        "private_subnets": [
            "subnet-xxxxxxxxxxxxxxxxx",
            "subnet-xxxxxxxxxxxxxxxxx",
            "subnet-xxxxxxxxxxxxxxxxx",
            "subnet-xxxxxxxxxxxxxxxxx"
        ],
        "public_subnets": [
            "subnet-xxxxxxxxxxxxxxxxx",
            "subnet-xxxxxxxxxxxxxxxxx",
            "subnet-xxxxxxxxxxxxxxxxx",
            "subnet-xxxxxxxxxxxxxxxxx"
        ],
        "vpc_cidr_block": "10.0.0.0/16",
        "vpc_id": "vpc-xxxxxxxxxxxxxxxxx"
    },
    "vpc_id": "vpc-xxxxxxxxxxxxxxxxx",
    "zone_id": "ROUTE53_EXTERNAL_ZONE_ID",
    "external_zone_name": "external.dns.zone",
    "eks_cluster": {
        "cluster_arn": "arn:aws:eks:us-west-2:00000000000:cluster/my-eks-cluster",
        "cluster_ca": "...",
        "cluster_endpoint": "https://A1B2C3D4.gr7.us-west-2.eks.amazonaws.com",
        "cluster_id": "my-eks-cluster",
        "cluster_oidc_issuer_url": "https://oidc.eks.us-west-2.amazonaws.com/id/A1B2C3D4",
        "cluster_version": "1.27",
        "oidc_provider_arn": "arn:aws:iam::00000000000:oidc-provider/oidc.eks.us-west-2.amazonaws.com/id/A1B2C3D4",
        "worker_iam_role_name": "my-eks-cluster-eks-node-role-name",
        "worker_security_group": "my-eks-cluster-worker-security-group"
    },
    "dbs": {},
    "dynamo_locktable_name": "dynamo-locktable-name",
    "ecrs": {},
    "hapi_config": {
        "assume_role_arn": "arn:aws:iam::00000000000:role/tfe-si",
        "base_url": "https://hapi.external.dns.zone",
        "kms_key_id": "kms-key-id",
        "oidc_authz_id": "oidc-authz-id",
        "oidc_issuer": "oidc-issuer",
        "scope": "happy"
    },
    "oidc_config": {
        "client_id": "xxxxxxxxxxxxxxxxx",
        "client_secret": "yyyyyyyyyyyyyyyyyy",
        "config_uri": "https://xxxxxxxxxxxxxxxxx:[email protected]/oauth2/",
        "idp_url": "my.okta.com"
    },
    "tags": {
        "env": "rdev",
        "owned_by": "happy"
    },
    "tfe": {
        "org": "happy",
        "url": "https://app.terraform.io"
    }
}

Substitute the values with the ones appropriate to your setup. hapi_config and oidc_confug sections are optional.

Create a happy namespace (happy-rdev) and apply the integration secret into it:

kubectl create ns happy-rdev
kubectl create secret generic integration-secret --from-file=integration_secret=./integration-secret.json -n happy-rdev

Install

Install happy:

MacOS

brew tap chanzuckerberg/tap
brew install happy

Alternatively, you can install hvm and install happy using hvm:

brew tap chanzuckerberg/tap
brew install hvm
hvm install chanzuckerberg happy <version>
hvm set-default chanzuckerberg happy <version>

Linux

Binaries are available on the releases page. Download one for your architecture, put it in your path and make it executable.

Instructions on downloading the binary:

1. Go here: https://github.com/chanzuckerberg/happy/releases to find which version of happy you want.
2. Run curl -s https://raw.githubusercontent.com/chanzuckerberg/happy/master/download.sh | bash -s -- -b HAPPY_PATH VERSION
   1. HAPPY_PATH is the directory where you want to install happy
   2. VERSION is the release you want
3. To verify you installed the desired version, you can run happy version.

Getting started

Setting up a brand new application

Create a folder called myapp

mkdir myapp
cd myapp

Bootstrap the Happy application:

happy bootstrap --force

Answer the prompts:

• What would you like to name this application?: myapp.
• Your application will be deployed to multiple environments. Which environments would you like to deploy to?: rdev
• Which aws profile do you want to use in rdev?: select the appropriate aws configuration profile
• Which aws region should we use in rdev?: select the aws region with the EKS cluster
• Which EKS cluster should we use in rdev?: select the cluster you will be deploying to
• Which happy namespace should we use in rdev?: select the namespace that has a Happy environment configured
• Would you like to use dockerfile ./Dockerfile as a service in your stack?: Y
• What would you like to name the service for ./Dockerfile?: myapp
• What kind of service is myapp?: EXTERNAL
• Which port does service myapp listen on?: use a port number other than 80 or 443
• Which uri does myapp respond on?: /
• File /tmp/myapp/docker-compose.yml already exists. Would you like to overwrite it, save a backup, or skip the change?: overwrite (if prompted)

At this point, your folder structure looks like

.
├── .happy
│   ├── config.json
│   └── terraform
│       └── envs
│           └── rdev
│               ├── main.tf
│               ├── outputs.tf
│               ├── providers.tf
│               ├── variables.tf
│               └── versions.tf
├── Dockerfile
└── docker-compose.yml

Happy configuration is blended from three sources: config.json for environment and application structure setup; main.tf to wire the terraform code and provide baseline parameters, and docker-compose.yaml to indicate where relevant Dockerfile files are located. Multiple environments (think dev, staging, and prod) can be defined with unique configuration settings.

Let's create a new stack: happy create myapp-stack, not that we have namespaced the stack name, as stack names are unique in the entire environment due to DNS constraints. Once the stack is created, happy will display a list of endpoints:

[INFO]: service_endpoints: {
	"EXTERNAL_MYAPP_ENDPOINT": "https://myapp-stack.<PARENT-DNS-ZONE-NAME>",
	"PRIVATE_MYAPP_ENDPOINT": "http://myapp-myapp.<NAMESPACE>.svc.cluster.local:80"
}

EXTERNAL_MYAPP_ENDPOINT is accessible from your browser. PRIVATE_MYAPP_ENDPOINT is accessible by other applications on running on the same cluster (if Linkerd is enabled, you can have granular controls over who can connect to it). Try curl-ing the EXTERNAL_MYAPP_ENDPOINT. You will get a default nginx response.

Now, list stacks out: happy list, you will get a human readable output. For machine-readable output, run happy list --output json. If you intend to update the application after changes were made, run happy update myapp-stack. Close the session by deleting the stack: happy update myapp-stack.

Sample apps

Clone this repo:

git clone https://github.com/chanzuckerberg/happy.git

Navigate to an example app and try happy out:

cd examples/typical_app
happy list
happy create mystack
happy update mystack
happy delete mystack

Integration Testing

All release-please pull requests automatically trigger an Integration Test workflow, which has to complete successfully for Happy to be released. This workflow does not run automatically on feature pull requests. If you wish to run an integration test on a pull request, add a happy:integration-test label to it.

Contributing

This project adheres to the Contributor Covenant code of conduct. By participating, you are expected to uphold this code. Please report unacceptable behavior to [email protected]. //