Deploying an app based on a Docker image

Page last updated:

You can use the Cloud Foundry Command Line Interface (cf CLI) to push an app with a new or updated Docker image. Cloud Foundry Application Runtime then uses the Docker image to create containers for the app.

For an explanation of how Docker works in Cloud Foundry, see Using Docker in Cloud Foundry.

Requirements

To push an app based on a Docker image, you need:

  • A Cloud Foundry deployment with Docker support activated. To enable Docker support, see the Enable Docker section of the Using Docker in Cloud Foundry topic.

  • A Docker image that meets the following requirements:

    • The Docker image must contain an /etc/passwd file with an entry for the root user. In addition, the home directory and the shell for that root user must be present in the image file system.
    • The total size of the Docker image file system layers must not exceed the disk quota for the app. The maximum disk allocation for apps is set by the Cloud Controller. The default maximum disk quota is 2048 MB per app.

      If the total size of the Docker image file system layers exceeds the disk quota, the app instances do not start.

  • The location of the Docker image on Docker Hub or another Docker registry.

  • A registry that supports the Docker Registry HTTP API V2 and presents a valid certificate for HTTPS traffic. For more information, see the Docker Registry HTTP API V2 spec in the Docker documentation.

Requirement for cf ssh support

To log in to your application container using the cf ssh command, you must make a shell such as sh or bash available in the container.

The SSH server in the container looks for the following executables in absolute locations or the PATH environment variable:

  • /bin/bash

  • /usr/local/bin/bash

  • /bin/sh

  • bash

  • sh

Benefits of specifying tags

To make your application container consistent after platform updates and code changes, specify a tag when you push your Docker image. Otherwise, the platform applies the latest tag without respecting changes to PORT or ENTRYPOINT.

If you push your Docker image without specifying a tag, you must run cf restage for the changes to take effect.

Port configuration

By default, apps listen for connections on the port specified in the PORT environment variable for the app. Cloud Foundry allocates this value dynamically.

When configuring a Docker image for Cloud Foundry, you can control the exposed port and the corresponding value of PORT by specifying the EXPOSE directive in the image Dockerfile. If you specify the EXPOSE directive, then the corresponding app pushed to Cloud Foundry listens on that exposed port. For example, if you set EXPOSE to 7070, then the app listens for connections on port 7070.

If you do not specify a port in the EXPOSE directive, then the app listens on the value of the PORT environment variable as determined by Cloud Foundry.

If you set the PORT environment variable via an ENV directive in a Dockerfile, Cloud Foundry overrides the value with the system-determined value.

Cloud Foundry supports only one exposed port on the image.

For more information about the PORT environment variable, see the PORT section of Cloud Foundry Environment variables. For more information about the EXPOSE directive, see the EXPOSE in the Docker documentation.

Start command

By default, Docker uses the start command specified by the CMD and/or ENTRYPOINT directives in the Docker image. You can override the start command either by using the -c flag on cf push or by setting the command property in a manifest file.

Push a Docker image from a registry

Cloud Foundry supports pushing apps from container registries such as Docker Hub, Google Container Registry (GCR), and Amazon Elastic Container Registry (ECR).

How you run cf push with apps stored in container registries depends on which registry you use and how it authenticates requests for the container image.

The following sections explain how to push apps under different container registry scenarios.

Docker Hub

To deploy a Docker image from a Docker Hub repository, run:

cf push APP-NAME --docker-image REPO/IMAGE:TAG

Where:

  • APP-NAME is the name to give the pushed app on Cloud Foundry.

  • REPO is the name of the repository where the image is stored.

  • IMAGE is the name of the app image on Docker Hub.

  • (Optional, but recommended) TAG is the tag or version for the image.

For example, the following command pushes the your-image image from Docker Hub to a Cloud Foundry app:

cf push your-app --docker-image cloudfoundry/your-image

Private container registry without authentication

As an alternative to Docker Hub, you can use any Docker image registry that presents a valid certificate for HTTPS traffic, such as a company-internal Docker registry.

To push an app as a Docker image using a specified Docker registry, run:

cf push APP-NAME --docker-image YOUR-PRIVATE-REGISTRY.DOMAIN:PORT/REPO/IMAGE:TAG

Where:

  • APP-NAME is the name to give the pushed app on Cloud Foundry.

  • YOUR-PRIVATE-REGISTRY.DOMAIN is the path to the Docker registry.

  • PORT is the port where the registry serves traffic.

  • REPO is the name of the repository where the image is stored.

  • IMAGE is the name of the app image being pushed.

  • (Optional, but recommended) TAG is the tag or version for the image.

For example, the following command pushes the v2 version of the your-image image from the your-repo repository of the internal-registry.example.com registry on port 5000:

cf push your-app --docker-image internal-registry.example.com:5000/your-repo/your-image:v2

Private container registry with basic authentication

Many Docker registries control access to Docker images by authenticating with a user name and password.

To push an app as a Docker image from a registry that uses basic user name and password authentication, run:

CF_DOCKER_PASSWORD=YOUR-PASSWORD cf push APP-NAME --docker-image REPO/IMAGE:TAG --docker-username USER

Where:

  • YOUR-PASSWORD is the password to use for authentication with the Docker registry.

    • Setting CF_DOCKER_PASSWORD prepended to the cf push --docker-image makes the value temporary, which is more secure than setting the environment variable indefinitely with export.
  • APP-NAME is the name to give the pushed app on Cloud Foundry.

  • REPO is the repository where the image is stored.

    • For Docker Hub, this is just the repository name.
    • For a private registry, this includes the registry address and port, as described in Push a Docker image from a private registry, in the format YOUR-PRIVATE-REGISTRY.DOMAIN:PORT/REPO.
  • IMAGE is the name of the app image being pushed.

  • (Optional, but recommended) TAG is the tag or version for the image.

  • USER is the user name to use for authentication with the registry.

If container registry credentials change, you have two options for an update: either you push the app with the new credentials or you update the latest package with the new credentials using PATCH /v3/packages and then restage your app. Apps require access to the container registry when starting. If you do not update the app with the new credentials, Cloud Foundry fails to start the app. When you rotate container credentials, Cloud Foundry recommends using a set of two credentials, where the old credentials can be deactivated after all apps are pushed with the new credentials.

Amazon Elastic Container Registry (ECR)

Cloud Foundry supports pushing apps from images hosted on Amazon Web Services ECR, which authenticates with temporary password tokens.

To push an app as a Docker image from ECR, run:

CF_DOCKER_PASSWORD=AWS-SECRET-ACCESS-KEY cf push APP-NAME --docker-image REPO/IMAGE:TAG --docker-username AWS-ACCESS-KEY-ID

Where:

  • AWS-SECRET-ACCESS-KEY is the AWS Secret Access Key for the IAM user accessing the ECR registry.

    • Setting CF_DOCKER_PASSWORD prepended to the cf push --docker-image makes the value temporary, which is more secure than setting the environment variable indefinitely with export.
  • APP-NAME is the name to give the pushed app on Cloud Foundry.

  • REPO is the ECR repository containing the image being pushed.

  • IMAGE is the name of the app image being pushed.

  • (Optional, but recommended) TAG is the tag or version for the image.

  • AWS-ACCESS-KEY-ID is the AWS Access Key ID for the IAM user accessing the ECR registry.

Running cf push with an ECR registry triggers Cloud Foundry to:

  1. Use the AWS Secret Access Key and Access Key ID to retrieve the temporary ECR user name and password.

  2. Use the temporary tokens to retrieve the image.

Google Container Registry (GCR)

Cloud Foundry Application Runtime supports pushing apps from images hosted on Google Container Registry (GCR) service. This feature requires that you use JSON key-based authentication. For more information about JSON key authentication, see the Google Cloud documentation.

Step 1: Authenticate with GCR

To authenticate with GCR, you must create a JSON key file and associate it with your project.

To create a JSON key file and associate it with your project:

  1. Create a GCP service account. To create a GCP service account, see the Google Cloud documentation. Run:

    gcloud iam service-accounts create YOUR-ACCOUNT --display-name "YOUR-DISPLAY-NAME"
    

    Where:

    • YOUR-ACCOUNT is the name of your service account.
    • YOUR-DISPLAY-NAME is the display name of your service account.

  2. Set your project ID by running:

    gcloud config set project YOUR-PROJECT-ID
    

    Where YOUR-PROJECT-ID is your project ID.

  3. Create a JSON key file and associate it with the service account by running:

    gcloud iam service-accounts keys create key.json --iam-account=YOUR-ACCOUNT@YOUR-PROJECT-ID.iam.gserviceaccount.com
    

    Where:

    • YOUR-ACCOUNT is the name of your service account.
    • YOUR-PROJECT-ID is your project ID.

  4. Add the IAM policy binding for your project and service account by running:

    gcloud projects add-iam-policy-binding YOUR-PROJECT --member serviceAccount:YOUR-ACCOUNT@YOUR-PROJECT-ID.iam.gserviceaccount.com --role roles/storage.objectViewer
    

    Where:

    • YOUR-PROJECT is the name of your project.
    • YOUR-ACCOUNT is the name of your service account.
    • YOUR-PROJECT-ID is your project ID.

Step 2: Deploy the GCP image

To deploy your GCR image using the cf CLI, run:

CF_DOCKER_PASSWORD="$(cat key.json)" cf push APP-NAME --docker-image docker://YOUR-REGISTRY-URL/YOUR-PROJECT/YOUR-IMAGE-NAME --docker-username _json_key`

Where:

  • APP-NAME is the name of the app being pushed.

  • YOUR-REGISTRY-URL is the URL of your registry.

  • YOUR-PROJECT is the name of your project.

  • YOUR-IMAGE-NAME is the name of your image.

  • The key.json file must point to the file you created earlier.

For information about specifying YOUR-REGISTRY-URL, see Pushing and Pulling Images in the Google Cloud documentation.

Docker volume support

You can use volume services with Docker apps. For more information about enabling volume support, see Using an external file system (volume services).

View the source for this page in GitHub