Secrets management in Docker is a critical security concern for any business. When using Docker containers, it is essential to keep sensitive data such as passwords, API keys, and other credentials secure.

This blog post will discuss some best practices for managing secrets in Docker, including how to store them securely and minimize their exposure. We will explore multiple solutions: using Docker Secrets with Docker Swarm, Docker Compose, or Mozilla SOPS. Feel free to choose what’s more appropriate to your use case. But most importantly is to remember to never hard-code your Docker secrets in plaintext in your Dockerfile!

Following these guidelines ensures that your organization's sensitive information remains safe even when running containerized services.

4 Ways to Store & Manage Secrets in Docker

Using Docker Secrets & Docker Swarm

Docker Secrets and Docker Swarm are two official and complimentary tools allowing to securely manage secrets when running containerized services.

Docker Secrets provides a secure mechanism for storing and retrieving secrets from the system without exposing them in plaintext. It enables users to keep their credentials safe by encrypting the data with a unique key before passing it to the system.

Docker Swarm is a powerful tool for managing clusters of nodes for distributed applications. It provides an effective means of deploying containerized applications at scale. With this tool, you can easily manage multiple nodes within a cluster and automatically distribute workloads among them. This helps ensure that your application has enough resources available at all times, even during peak usage periods or unexpected traffic spikes.

Together, these two tools provide an effective way to ensure that your organization's sensitive information remains safe despite ever-evolving security needs.

Let’s see how create and manage an example secret.

Creating a Secret

To create a secret, we need to first initialize Docker Swarm. You can do so using the following command:

docker swarm init

Once the service is initialized, we can use the docker secret create command to create the secret:

ssh-keygen -t rsa -b 4096 -N "" -f mykey
docker secret create my_key mykey
rm mykey

In these commands, we first create an SSH key using the ssh-keygen command and write it to mykey. Then, we use the docker secret command to generate the secret. Ensure that you delete the mykey file to avoid any security risks.

You can use the following command to confirm that the secret is created successfully:

docker secret ls

We can now use this secret in our Docker containers. One way is to pass this secret with –secret flag when creating a service.

docker service  create --name mongodb --secret my_mongodb_secret redis:latest

We can also pass this secret to docker-compose.yml file. Let’s take a look at an example file:

version: '3.7'
services:
  myapp:
    image: mydummyapp:latest
    secrets:
      - my_secret_key
    
secrets:
  my_secret_key:
    external: true

In the example compose file, the secrets section defines a secret named my_secret_key (discussed earlier). The myapp service definition specifies that it requires my_secret_key, and Docker will automatically mount it as a file at /run/secrets/my_secret_key in the container.

Using Docker Compose

Docker Compose is a powerful tool for defining and running multi-container applications with Docker. A stack is defined by a docker-compose file allowing you to define and configure the services that make up your application, including their environment variables, networks, ports, and volumes. With Docker Compose, it is easy to set up an application in a single configuration file and deploy it quickly and consistently across multiple environments.

Docker Compose provides an effective solution for managing secrets for organizations handling sensitive data such as passwords or API keys. You can read your secrets from an external file (like a TXT file). But be careful not to commit this file with your code!

version: '3.7'
services:
  myapp:
    image: myapp:latest
    secrets:
      - my_secret
secrets:
  my_secret:
    file: ./my_secret.txt

Using a Sidecar Container

A typical strategy for maintaining and storing secrets in a Docker environment is to use sidecar containers. Secrets can be sent to the main application container via the sidecar container, which can also operate a secrets manager or another secure service.

Let’s understand this using a Hashicorp Vault sidecar for a MongoDB container:

• First, create a Docker Compose (docker-compose.yml) file with two services: mongo and secrets.
• In the secrets service, use an image containing your chosen secret management tool, such as a vault.
• Mount a volume from the secrets container to the mongo container so the mongo container can access the secrets stored in the secrets container.
• In the mongo service, use environment variables to set the credentials for the MongoDB database, and reference the secrets stored in the mounted volume.

Here is the example compose file:

version: '3.7'

services:
  mongo:
    image: mongo
    volumes:
      - secrets:/run/secrets
    environment:
      MONGO_INITDB_ROOT_USERNAME_FILE: /run/secrets/mongo-root-username
      MONGO_INITDB_ROOT_PASSWORD_FILE: /run/secrets/mongo-root-password
  secrets:
    image: vault
    volumes:
      - secrets:/secrets
    command: ["vault", "server", "-dev", "-dev-root-token-id=myroot"]
    ports:
      - "8200:8200"
      
volumes:
  secrets:

Using Mozilla SOPS

Finally, your Docker host could decrypt a secret locally and pass that value to a Docker container as a temporary environment variable. One such tool for this is Mozilla SOPS. This tutorial assumes that SOPS is already installed on the Docker host. If you need an introduction to SOPS, read a primer here.

The command below shows the decrypted contents of an encrypted file, secret.enc.json. This file is safe to store on disk because it has been encrypted. SOPS can decrypt it because it has access to the encryption key.

$ sops decrypt secret.enc.json
{
        "PASSWORD_1: "SuperSecretPassword",
        "PASSWORD_2": "AnotherSecretPassword"
} 

The exec-env feature of SOPS allows us to extract key-value pairs from an encrypted file that is structured as YAML or JSON, and pass these as environment variables to a child process. The following command shows an example of launching a Docker container from the command line in this way.

sops exec-env secret.enc.json 'docker run --rm -it -e PASSWORD_1=$PASSWORD_1 bash -c "echo $PASSWORD_1"'

Of course, we could use this alongside Docker Compose, similar to previous examples. Here is an example Compose file, docker-compose.yml:

services:
  env_printer:
    image: alpine
    command: [ "sh", "-c", "echo The secret is: $PASSWORD_1" ]
    environment:
      - PASSWORD_1

To pass environment variables to Docker Compose, but not other processes, run the following command:

# decrypt our file with exec-env and run docker compose which can use these temporary env vars.
sops exec-env secret.enc.json 'docker compose up'

Importantly, this environment variable is available only to the process that runs Docker Compose, and not within the shell in which the sops command was run.

Scan for secrets in your Docker images

Hard coding secrets in Docker is a significant security risk, making them vulnerable to attackers. We have seen different best practices to avoid hard-coding secrets in plaintext in your Docker images, but security doesn't stop there.

You should also scan your images for secrets.

All Dockerfiles start with a FROM directive that defines the base image. It's important to understand that when you use a base image, especially from a public registry like Docker Hub, you are pulling external code that may contain hardcoded secrets. More information is exposed than visible in your single Dockerfile. Indeed, it's possible to retrieve a plaintext secret hard-coded in a previous layer starting from your image.

In fact, many public Docker images are concerned: in 2021, we estimated that 7% of the Docker Hub images contained at least one secret.

Secrets exposed in Docker images: Hunting for secrets in Docker Hub

In this article, we will explain why Docker images can contain sensitive information and give some examples of the type of secrets we found in public Docker images. Finally, we will compare our results to the ones we have with source code scanning.

GitGuardian Blog - Automated Secrets DetectionGuardians

Fortunately, you can easily detect them with ggshield (GitGuardian CLI). For example:

ggshield secret scan docker ubuntu:22.04

Conclusion

To sum up, managing secrets in Docker is a crucial part of preserving the security of your containerized apps. Docker includes several built-in tools for maintaining secrets, such as Docker Secrets and Docker Compose files.

Additionally, organizations can use third-party solutions like HashiCorp Vault and Mozilla SOPS to manage secrets in Docker. These technologies offer extra capabilities like access control, encryption, and audit logging to strengthen the security of your secret management.

Finally, finding and limiting accidental or unintended exposure of sensitive information is crucial to handling secrets in Docker. Companies are invited to use secret scanning tools such as GitGuardian to scan the Docker images built in their CI/CD pipelines as mitigation to prevent supply-chain attacks.

If you want to know more about Docker security, we also summarized some of the best practices in a cheat sheet.

We hope this blog post has provided you with a better understanding of how to manage secrets in Docker and keep your applications secure.

It is part of a series on secrets management with popular technologies, have a look!

How to Handle Secrets in Jenkins

DevOps engineers must handle secrets with care. In this series, we summarize best practices for leveraging secrets with your everyday tools.

GitGuardian Blog - Automated Secrets DetectionGuest Expert

How to Handle Secrets in Python

DevOps engineers must handle secrets with care. In this series, we summarize best practices for leveraging secrets with your everyday tools.

GitGuardian Blog - Automated Secrets DetectionGuest Expert

How to Handle AWS Secrets

In this blog post, we’ll cover some best practices for managing AWS secrets when using the AWS SDK in Python.

GitGuardian Blog - Automated Secrets DetectionGuest Expert

This article is a guest post. Views and opinions expressed in this publication are solely those of the author and do not reflect the official policy, position, or views of GitGuardian, The content is provided for informational purposes, GitGuardian assumes no responsibility for any errors, omissions, or outcomes resulting from the use of this information. Should you have any enquiry with regard to the content, please contact the author directly.