Amazon ECS

Learn how to deliver secrets to Amazon Elastic Container Service.

This guide will go over the steps needed to access secrets stored in Infisical from Amazon Elastic Container Service (ECS).

At a high level, the steps involve setting up an ECS task with a Infisical Agent as a sidecar container. This sidecar container uses Universal Auth to authenticate with Infisical to fetch secrets/access tokens. Once the secrets/access tokens are retrieved, they are then stored in a shared Amazon Elastic File System (EFS) volume. This volume is then made accessible to your application and all of its replicas.

This guide primarily focuses on integrating Infisical Cloud with Amazon ECS on AWS Fargate and Amazon EFS. However, the principles and steps can be adapted for use with any instance of Infisical (on premise or cloud) and different ECS launch configurations.

Prerequisites

This guide requires the following prerequisites:

• Infisical account
• Git installed
• Terraform v1.0 or later installed
• Access to AWS credentials
• Understanding of Infisical Agent

What we will deploy

For this demonstration, we’ll deploy the File Browser application on our ECS cluster. Although this guide focuses on File Browser, the principles outlined here can be applied to any application of your choice.

File Browser plays a key role in this context because it enables us to view all files attached to a specific volume. This feature is important for our demonstration, as it allows us to verify whether the Infisical agent is depositing the expected files into the designated file volume and if those files are accessible to the application.

Volumes that contain sensitive secrets should not be publicly accessible. The use of File Browser here is solely for demonstration and verification purposes.

Configure Authentication with Infisical

In order for the Infisical agent to fetch credentials from Infisical, we’ll first need to authenticate with Infisical. While Infisical supports various authentication methods, this guide focuses on using Universal Auth to authenticate the agent with Infisical.

Follow the documentation to configure and generate a client id and client secret with Universal auth here. Make sure to save these credentials somewhere handy because you’ll need them soon.

Clone guide assets repository

To help you quickly deploy the example application, please clone the guide assets from this Github repository. This repository contains assets for all Infisical guides. The content for this guide can be found within a sub directory called aws-ecs-with-agent. The guide will assume that aws-ecs-with-agent is your working directory going forward.

Deploy example application

Before we can deploy our full application and its related infrastructure with Terraform, we’ll need to first configure our Infisical agent.

Agent configuration overview

The agent config file defines what authentication method will be used when connecting with Infisical along with where the fetched secrets/access tokens should be saved to.

Since the Infisical agent will be deployed as a sidecar, the agent configuration file and any secret template files will need to be encoded in base64. This encoding step is necessary as it allows these files to be added into our Terraform configuration file without needing to upload them first.

Secret template file

The Infisical agent accepts one or more optional template files. If provided, the agent will fetch secrets using the set authentication method and format the fetched secrets according to the given template file.

For demonstration purposes, we will create the following secret template file. This template will transform our secrets from Infisical project with the ID 62fd92aa8b63973fee23dec7, in the dev environment, and secrets located in the path /, into a KEY=VALUE format.

Remember to update the project id, environment slug and secret path to one that exists within your Infisical project

secrets.template

{{- with secret "62fd92aa8b63973fee23dec7" "dev" "/" }}
{{- range . }}
{{ .Key }}={{ .Value }}
{{- end }}
{{- end }}

Next, we need encode this template file in base64 so it can be set in the agent configuration file.

cat secrets.template | base64 
Cnt7LSB3aXRoIHNlY3JldCAiMWVkMjk2MWQtNDM5NS00MmNlLTlkNzQtYjk2ZGQwYmYzMDg0IiAiZGV2IiAiLyIgfX0Ke3stIHJhbmdlIC4gfX0Ke3sgLktleSB9fT17eyAuVmFsdWUgfX0Ke3stIGVuZCB9fQp7ey0gZW5kIH19

Full agent configuration file

This agent config file will connect with Infisical Cloud using Universal Auth and deposit access tokens at path /infisical-agent/access-token and render secrets to file /infisical-agent/secrets.

You’ll notice that instead of passing the path to the secret template file as we normally would, we set the base64 encoded template from the previous step under base64-template-content property.

agent-config.yaml

infisical:
  address: https://app.infisical.com
  exit-after-auth: true
auth:
  type: universal-auth
  config:
    remove_client_secret_on_read: false
sinks:
  - type: file
    config:
      path: /infisical-agent/access-token
templates:
  - base64-template-content: Cnt7LSB3aXRoIHNlY3JldCAiMWVkMjk2MWQtNDM5NS00MmNlLTlkNzQtYjk2ZGQwYmYzMDg0IiAiZGV2IiAiLyIgfX0Ke3stIHJhbmdlIC4gfX0Ke3sgLktleSB9fT17eyAuVmFsdWUgfX0Ke3stIGVuZCB9fQp7ey0gZW5kIH19
    destination-path: /infisical-agent/secrets

Again, we’ll need to encode the full configuration file in base64 so it can be easily delivered via Terraform.

cat agent-config.yaml | base64 
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

Add auth credentials & agent config

With the base64 encoded agent config file and Universal Auth credentials in hand, it’s time to assign them as values in our Terraform config file.

To configure these values, navigate to the ecs.tf file in your preferred code editor and assign values to auth_client_id, auth_client_secret, and agent_config.

ecs.tf

...snip...
data "template_file" "cb_app" {
  template = file("./templates/ecs/cb_app.json.tpl")

  vars = {
    app_image          = var.app_image
    sidecar_image      = var.sidecar_image
    app_port           = var.app_port
    fargate_cpu        = var.fargate_cpu
    fargate_memory     = var.fargate_memory
    aws_region         = var.aws_region
    auth_client_id     = "<paste-client-id-string>"
    auth_client_secret = "<paset-client-secret-string>"
    agent_config       = "<paste-base64-encoded-agent-config-string>"
  }
}
...snip...

To keep this guide simple, auth_client_id, auth_client_secret have been added directly into the ECS container definition. However, in production, you should securely fetch these values from AWS Secrets Manager or AWS Parameter store and feed them directly to agent sidecar.

After these values have been set, they will be passed to the Infisical agent during startup through environment variables, as configured in the infisical-sidecar container below.

templates/ecs/cb_app.json.tpl

[
...snip...
  {
    "name": "infisical-sidecar",
    "image": "${sidecar_image}",
    "cpu": 1024,
    "memory": 1024,
    "networkMode": "bridge",
    "command": ["agent"],
    "essential": false,
    "logConfiguration": {
        "logDriver": "awslogs",
        "options": {
          "awslogs-group": "/ecs/agent",
          "awslogs-region": "${aws_region}",
          "awslogs-stream-prefix": "ecs"
        }
    },
    "healthCheck": {
        "command": ["CMD-SHELL", "agent", "--help"],
        "interval": 30,
        "timeout": 5,
        "retries": 3,
        "startPeriod": 0
    },
    "environment": [
      {
        "name": "INFISICAL_UNIVERSAL_AUTH_CLIENT_ID",
        "value": "${auth_client_id}"
      },
      {
        "name": "INFISICAL_UNIVERSAL_CLIENT_SECRET",
        "value": "${auth_client_secret}"
      },
      {
        "name": "INFISICAL_AGENT_CONFIG_BASE64",
        "value": "${agent_config}"
      }
    ],
    "mountPoints": [
      {
          "containerPath": "/infisical-agent",
          "sourceVolume": "infisical-efs"
      }
    ]
  }
]

In the above container definition, you’ll notice that that the Infisical agent has a mountPoints defined. This mount point is referencing to the already configured EFS volume as shown below. containerPath is set to /infisical-agent because that is that the folder we have instructed the agent to deposit the credentials to.

terraform/efs.tf

resource "aws_efs_file_system" "infisical_efs" {
  tags = {
    Name = "INFISICAL-ECS-EFS"
  }
}

resource "aws_efs_mount_target" "mount" {
  count           = length(aws_subnet.private.*.id)
  file_system_id  = aws_efs_file_system.infisical_efs.id
  subnet_id       = aws_subnet.private[count.index].id
  security_groups = [aws_security_group.efs_sg.id]
}

Configure AWS credentials

Because we’ll be deploying the example file browser application to AWS via Terraform, you will need to obtain a set of AWS Access Key and Secret Key. Once you have generated these credentials, export them to your terminal.

1. Export the AWS Access Key ID:

   export AWS_ACCESS_KEY_ID=<your AWS access key ID>
   

2. Export the AWS Secret Access Key:

   export AWS_SECRET_ACCESS_KEY=<your AWS secret access key>
   

Deploy terraform configuration

With the agent’s sidecar configuration complete, we can now deploy our changes to AWS via Terraform.

1. Change your directory to terraform

cd terraform

2. Initialize Terraform

$ terraform init 

3. Preview resources that will be created

$ terraform plan

4. Trigger resource creation

$ terraform apply 

Do you want to perform these actions?
  Terraform will perform the actions described above.
  Only 'yes' will be accepted to approve.

  Enter a value: yes

Apply complete! Resources: 1 added, 1 changed, 1 destroyed.

Outputs:

alb_hostname = "cb-load-balancer-1675475779.us-east-1.elb.amazonaws.com:8080"

Once the resources have been successfully deployed, Terrafrom will output the host address where the file browser application will be accessible. It may take a few minutes for the application to become fully ready.

Verify secrets/tokens in EFS volume

To verify that the agent is depositing access tokens and rendering secrets to the paths specified in the agent config, navigate to the web address from the previous step. Once you visit the address, you’ll be prompted to login. Enter the credentials shown below.

Since our EFS volume is mounted to the path of the file browser application, we should see the access token and rendered secret file we defined via the agent config file.

As expected, two files are present: access-token and secrets. The access-token file should hold a valid Bearer token, which can be used to make HTTP requests to Infisical. The secrets file should contain secrets, formatted according to the specifications in our secret template file (presented in key=value format).