Cloud Controller Blobstore Configuration

Page last updated:

This topic explains how to configure a blobstore for the Cloud Controller.

Overview

The cf-deployment GitHub repository provides ops files with several common blobstore configurations. By default, cf-deployment uses the WebDAV blobstore, and no additional ops files are needed. If you want to configure your blobstore manually, see the topics below for guidance.

The Cloud Controller has four types of objects that need to be stored in a blobstore: buildpacks, droplets, packages, and resource_pool. By default, the blobstore configuration uses the Fog Ruby gem, but can also use the WebDAV protocol.

This document describes the following common blobstore configurations:

Fog with AWS Credentials

To use Fog blobstores with AWS credentials, do the following:

  1. Insert the following configuration into your manifest under properties.cc:

    cc:
      buildpacks:
        blobstore_type: fog
        buildpack_directory_key: YOUR-AWS-BUILDPACK-BUCKET
        fog_connection: &fog_connection
          aws_access_key_id: AWS-ACCESS-KEY
          aws_secret_access_key: AWS-SECRET-ACCESS-KEY
          provider: AWS
          region: us-east-1
      droplets:
        blobstore_type: fog
        droplet_directory_key: YOUR-AWS-DROPLET-BUCKET
        fog_connection: *fog_connection
      packages:
        blobstore_type: fog
        app_package_directory_key: YOUR-AWS-PACKAGE-BUCKET
        fog_connection: *fog_connection
      resource_pool:
        blobstore_type: fog
        resource_directory_key: YOUR-AWS-RESOURCE-BUCKET
        fog_connection: *fog_connection
    
  2. Replace AWS-ACCESS-KEY and AWS-SECRET-ACCESS-KEY with your AWS credentials.

  3. Replace YOUR-AWS-BUILDPACK-BUCKET, YOUR-AWS-DROPLET-BUCKET, YOUR-AWS-PACKAGE-BUCKET, and YOUR-AWS-RESOURCE-BUCKET with the names of your AWS buckets. Do not use periods (.) in your AWS bucket names. In the AWS console, you must assign your credentials an IAM policy that allows all S3 actions on all of these buckets.

  4. (Optional) Provide additional configuration through the fog_connection hash, which is passed through to the Fog gem.

Fog with AWS Server-Side Encryption

AWS S3 offers Server-Side Encryption at rest. For more information, see Protecting Data Using Server-Side Encryption.

AWS SSE-S3 blobstore encryption

  1. Insert the following configuration into your manifest under properties.cc:

      cc:
        buildpacks:
          blobstore_type: fog
          buildpack_directory_key: YOUR-AWS-BUILDPACK-BUCKET
          fog_connection: &fog_connection
            aws_access_key_id: AWS-ACCESS-KEY
            aws_secret_access_key: AWS-SECRET-ACCESS-KEY
            provider: AWS
            region: us-east-1
          fog_aws_storage_options: &fog_aws_storage_options
            encryption: 'AES256'
        droplets:
          blobstore_type: fog
          droplet_directory_key: YOUR-AWS-DROPLET-BUCKET
          fog_connection: *fog_connection
          fog_aws_storage_options: *fog_aws_storage_options
        packages:
          blobstore_type: fog
          app_package_directory_key: YOUR-AWS-PACKAGE-BUCKET
          fog_connection: *fog_connection
          fog_aws_storage_options: *fog_aws_storage_options
        resource_pool:
          blobstore_type: fog
          resource_directory_key: YOUR-AWS-RESOURCE-BUCKET
          fog_connection: *fog_connection
          fog_aws_storage_options: *fog_aws_storage_options
    
  2. Replace AWS_ACCESS_KEY and AWS_SECRET_ACCESS_KEY with your AWS credentials.

  3. Replace YOUR-AWS-BUILDPACK-BUCKET, YOUR-AWS-DROPLET-BUCKET, YOUR-AWS-PACKAGE-BUCKET, and YOUR-AWS-RESOURCE-BUCKET with the names of your AWS buckets. Do not use periods (.) in your AWS bucket names. In the AWS console, you must assign your credentials an IAM policy that allows all S3 actions on all of these buckets.

  4. You can provide further configuration through the fog_connection hash, which is passed through to the Fog gem.

  5. fog_aws_storage_options takes a hash with the key encryption. Operators can set its value to a type of encryption algorithm. In the configuration information above, encryption is set to AES256 to enable AWS SSE-S3 encryption.

  6. You can provide further configuration through the fog_aws_storage_options hash, which is passed through to the Fog gem.

AWS SSE-KMS blobstore encryption

  1. Obtain your KMS Key ID. For information about managing KMS keys, see the AWS Key Management Service Getting Started guide.

  2. Insert the following configuration into your manifest under properties.cc:

      cc:
        buildpacks:
          blobstore_type: fog
          buildpack_directory_key: YOUR-AWS-BUILDPACK-BUCKET
          fog_connection: &fog_connection
            aws_access_key_id: AWS-ACCESS-KEY
            aws_secret_access_key: AWS-SECRET-ACCESS-KEY
            provider: AWS
            region: us-east-1
          fog_aws_storage_options: &fog_aws_storage_options
            encryption: 'aws:kms'
            x-amz-server-side-encryption-aws-kms-key-id: "YOUR-AWS-KMS-KEY-ID"
        droplets:
          blobstore_type: fog
          droplet_directory_key: YOUR-AWS-DROPLET-BUCKET
          fog_connection: *fog_connection
          fog_aws_storage_options: *fog_aws_storage_options
        packages:
          blobstore_type: fog
          app_package_directory_key: YOUR-AWS-PACKAGE-BUCKET
          fog_connection: *fog_connection
          fog_aws_storage_options: *fog_aws_storage_options
        resource_pool:
          blobstore_type: fog
          resource_directory_key: YOUR-AWS-RESOURCE-BUCKET
          fog_connection: *fog_connection
          fog_aws_storage_options: *fog_aws_storage_options
    
  3. Replace AWS-ACCESS-KEY and AWS-SECRET-ACCESS-KEY with your AWS credentials.

  4. Replace YOUR-AWS-BUILDPACK-BUCKET, YOUR-AWS-DROPLET-BUCKET, YOUR-AWS-PACKAGE-BUCKET, and YOUR-AWS-RESOURCE-BUCKET with the names of your AWS buckets. Do not use periods (.) in your AWS bucket names. In the AWS console, you must assign your credentials an IAM policy that allows all S3 actions on all of these buckets.

  5. You can provide further configuration through the fog_connection hash, which is passed through to the Fog gem.

  6. Replace YOUR-AWS-KMS-KEY-ID with your KMS Key ID.

  7. fog_aws_storage_options takes a hash with the key encryption. Operators can set its value to a type of encryption algorithm. In the configuration information above, encryption is set to aws:kms to enable AWS SSE-KMS encryption.

  8. You can provide further configuration through the fog_aws_storage_options hash, which is passed through to the Fog gem.

Fog with AWS IAM Instance Profiles

To configure Fog blobstores to use AWS IAM Instance Profiles, do the following:

  1. Configure an additional cloud-controller IAM role with the following policy to give access to the S3 buckets you plan to use:

    {
      "Version": "2012-10-17",
      "Statement": [{
        "Effect": "Allow",
        "Action": [ "s3:*" ],
        "Resource": [
          "arn:aws:s3:::YOUR-AWS-BUILDPACK-BUCKET",
          "arn:aws:s3:::YOUR-AWS-BUILDPACK-BUCKET/*",
          "arn:aws:s3:::YOUR-AWS-DROPLET-BUCKET",
          "arn:aws:s3:::YOUR-AWS-DROPLET-BUCKET/*",
          "arn:aws:s3:::YOUR-AWS-PACKAGE-BUCKET",
          "arn:aws:s3:::YOUR-AWS-PACKAGE-BUCKET/*",
          "arn:aws:s3:::YOUR-AWS-RESOURCE-BUCKET",
          "arn:aws:s3:::YOUR-AWS-RESOURCE-BUCKET/*",
        ]
      }]
    }
    

    Replace YOUR-AWS-BUILDPACK-BUCKET, YOUR-AWS-DROPLET-BUCKET, YOUR-AWS-PACKAGE-BUCKET, and YOUR-AWS-RESOURCE-BUCKET with the names of your AWS buckets. Do not use periods (.) in your AWS bucket names.

    If you use the AWS console, an IAM Role is automatically assigned to an IAM Instance Profile with the same name, cloud-controller. If you do not use the AWS console, you must create an IAM Instance Profile with a single assigned IAM Role. For more information, see Step 4: Create an IAM Instance Profile for Your Amazon EC2 Instances in the AWS documentation.

  2. In your BOSH cloud config, create a VM extension to add the IAM Instance Profile you created to VMs using the extension.

    vm_extensions:
    - cloud_properties:
        iam_instance_profile: cloud-controller
      name: cloud-controller-iam
    
  3. In your Cloud Foundry deployment manifest, use the cloud-controller-iam VM extension you created for the instance groups containing cloud_controller, cloud_controller_worker, and cloud_controller_clock, as in the example below:

    instance_groups:
    ...
    - name: api
      ...
      vm_extensions:
      - cloud-controller-iam
    ...
    - name: cc-worker
      ...
      vm_extensions:
      - cloud-controller-iam
    ...
    - name: scheduler
      ...
      vm_extensions:
      - cloud-controller-iam
    
  4. Insert the following configuration into your deployment manifest under properties.cc:

    cc:
      buildpacks:
        blobstore_type: fog
        buildpack_directory_key: YOUR-AWS-BUILDPACK-BUCKET
        fog_connection: &fog_connection
          provider: AWS
          region: us-east-1
          use_iam_profile: true
      droplets:
        blobstore_type: fog
        droplet_directory_key: YOUR-AWS-DROPLET-BUCKET
        fog_connection: *fog_connection
      packages:
        blobstore_type: fog
        app_package_directory_key: YOUR-AWS-PACKAGE-BUCKET
        fog_connection: *fog_connection
      resource_pool:
        blobstore_type: fog
        resource_directory_key: YOUR-AWS-RESOURCE-BUCKET
        fog_connection: *fog_connection
    

    Replace YOUR-AWS-BUILDPACK-BUCKET, YOUR-AWS-DROPLET-BUCKET, YOUR-AWS-PACKAGE-BUCKET, and YOUR-AWS-RESOURCE-BUCKET with the names of your AWS buckets. Do not use periods (.) in your AWS bucket names.

  5. (Optional) Provide other configuration with the fog_connection hash, which is passed through to the Fog gem.

Fog with Google Cloud Storage

To configure your blobstores to use Google Cloud Storage Interoperability credentials, do the following:

  1. Insert the following configuration into your manifest under properties.cc:

    cc:
      buildpacks:
        blobstore_type: fog
        buildpack_directory_key: YOUR-GCS-BUILDPACK-BUCKET
        fog_connection: &fog_connection
          provider: Google
          google_storage_access_key_id: GCS-ACCESS-KEY
          google_storage_secret_access_key: GCS-SECRET-ACCESS-KEY
      droplets:
        blobstore_type: fog
        droplet_directory_key: YOUR-GCS-DROPLET-BUCKET
        fog_connection: *fog_connection
      packages:
        blobstore_type: fog
        app_package_directory_key: YOUR-GCS-PACKAGE-BUCKET
        fog_connection: *fog_connection
      resource_pool:
        blobstore_type: fog
        resource_directory_key: YOUR-GCS-RESOURCE-BUCKET
    
  2. Create an Interoperability key using the GCP instructions.

  3. Replace GCS-ACCESS-KEY and GCS-SECRET-ACCESS-KEY with your Cloud Storage credentials.

  4. Replace YOUR-GCS-BUILDPACK-BUCKET, YOUR-GCS-DROPLET-BUCKET, YOUR-GCS-PACKAGE-BUCKET, and YOUR-GCS-RESOURCE-BUCKET with the names of your Cloud Storage buckets. Do not use periods (.) in your bucket names.

  5. You can provide further configuration through the fog_connection hash, which is passed through to the Fog gem.

Fog with Google Cloud Storage Service Accounts

To configure your blobstore to use Google Cloud Storage with a service account, do the following:

  1. Create a custom Cloud IAM role with the following permissions:

    storage.buckets.create
    storage.buckets.get
    storage.objects.create
    storage.objects.delete
    storage.objects.get
    storage.objects.list
    

    To create the custom role, follow the Creating and Managing Custom Roles instructions in the GCP documentation. This role will be used for blobstore permissions.

  2. Create a service account by following the Creating and Managing Service Accounts instructions in the GCP documentation and grant this service account the role you created in Step 1. For additional information about granting roles to an existing service account, see Granting Roles to a Service Account for Specific Resources in the GCP documentation.

  3. Create a service account JSON key for your service account by following the Creating and Managing Service Account Keys instructions in the GCP documentation. You need this key to configure your blobstore.

  4. Insert the following configuration in your manifest under properties.cc for the cloud_controller_ng job:

    cc:
      buildpacks: &buildpacks
        blobstore_type: fog
        buildpack_directory_key: YOUR-GCS-BUILDPACK-BUCKET
        fog_connection: &fog_connection
          provider: Google
          google_project: YOUR-GCS-PROJECT
          google_client_email: YOUR-GCS-SERVICE-ACCOUNT-EMAIL
          google_json_key_string: >
            {
              "type": "service_account",
              "project_id": "YOUR-GCS-PROJECT",
              ...
            }
      droplets: &droplets
        blobstore_type: fog
        droplet_directory_key: YOUR-GCS-DROPLET-BUCKET
        fog_connection: *fog_connection
      packages: &packages
        blobstore_type: fog
        app_package_directory_key: YOUR-GCS-PACKAGE-BUCKET
        fog_connection: *fog_connection
      resource_pool: &resource_pool
        blobstore_type: fog
        fog_connection: *fog_connection
        resource_directory_key: YOUR-GCS-RESOURCE-BUCKET
    

    Replace the placeholders as follows:

    • YOUR-GCS-PROJECT is the name of your GCP project.
    • YOUR-GCS-SERVICE-ACCOUNT-EMAIL is the email of your GCP service account.
    • YOUR-GCS-BUILDPACK-BUCKET, YOUR-GCS-DROPLET-BUCKET,
      YOUR-GCS-PACKAGE-BUCKET, and YOUR-GCS-RESOURCE-BUCKET are the names of your Cloud Storage buckets. Do not use periods in the bucket names.

    In google_json_key_string, provide your Cloud Storage credentials. Use spaces, not tabs, for indentation.

  5. To update the cloud_controller_worker and cloud_controller_clock jobs with the same blobstore configuration, insert the following for both jobs under properties.cc:

    cc:
      # note these reference YAML anchors declared in the previous step
      buildpacks: *buildpacks
      droplets: *droplets
      packages: *packages
      resource_pool: *resource_pool
    

Fog with Azure Storage

To configure your blobstores to use Azure Storage credentials, do the following:

  1. Insert the following configuration into your manifest under properties.cc:

    cc:
      buildpacks:
        blobstore_type: fog
        buildpack_directory_key: YOUR-AZURE-BUILDPACK-CONTAINER
        fog_connection: &fog_connection
          provider: AzureRM
          environment: AzureCloud
          azure_storage_account_name: YOUR-AZURE-STORAGE-ACCOUNT-NAME
          azure_storage_access_key: YOUR-AZURE-STORAGE-ACCESS-KEY
      droplets:
        blobstore_type: fog
        droplet_directory_key: YOUR-AZURE-DROPLET-CONTAINER
        fog_connection: *fog_connection
      packages:
        blobstore_type: fog
        app_package_directory_key: YOUR-AZURE-PACKAGE-CONTAINER
        fog_connection: *fog_connection
      resource_pool:
        blobstore_type: fog
        resource_directory_key: YOUR-AZURE-RESOURCE-CONTAINER
        fog_connection: *fog_connection
    
  2. Replace YOUR-AZURE-STORAGE-ACCOUNT-NAME and YOUR-AZURE-STORAGE-ACCESS-KEY with your Azure Storage credentials.

  3. Replace YOUR-AZURE-BUILDPACK-CONTAINER, YOUR-AZURE-DROPLET-CONTAINER, YOUR-AZURE-PACKAGE-CONTAINER, and YOUR-AZURE-RESOURCE-CONTAINER with the names of your Cloud Storage containers. Azure container names must consist of only lowercase alphanumeric characters and hyphens. See Azure’s storage name restrictions.

  4. You can provide further configuration through the fog_connection hash, which is passed through to the Fog gem.

Fog with Other S3 Compatible Stores

Using Fog blobstores with other S3 compatible stores, such as Minio or EMC Elastic Cloud Storage is similar to AWS, and uses the same Fog adapter, but requires slightly different configuration:

  1. Insert the following configuration into your manifest under properties.cc:

    cc:
      buildpacks:
        blobstore_type: fog
        buildpack_directory_key: YOUR-S3-BUILDPACK-BUCKET
        fog_connection: &fog_connection
          provider: AWS
          endpoint: S3-ENDPOINT
          aws_access_key_id: S3-ACCESS-KEY
          aws_secret_access_key: S3-SECRET-ACCESS-KEY
          aws_signature_version: '2'
          region: "''"
          path_style: true
      droplets:
        blobstore_type: fog
        droplet_directory_key: YOUR-S3-DROPLET-BUCKET
        fog_connection: *fog_connection
      packages:
        blobstore_type: fog
        app_package_directory_key: YOUR-S3-PACKAGE-BUCKET
        fog_connection: *fog_connection
      resource_pool:
        blobstore_type: fog
        resource_directory_key: YOUR-S3-RESOURCE-BUCKET
        fog_connection: *fog_connection
    
  2. Replace S3-ENDPOINT with the URL used to access your S3 API. This will typically look something like http://S3-NAMESPACE.HOST:9020 but may vary for your server or network.

  3. Replace S3-ACCESS-KEY and S3-SECRET-ACCESS-KEY with your S3 credentials. This key must have access to all S3 activities on the buckets you will specify below.

  4. Replace YOUR-S3-BUILDPACK-BUCKET, YOUR-S3-DROPLET-BUCKET, YOUR-S3-PACKAGE-BUCKET, and YOUR-S3-RESOURCE-BUCKET with the names of your S3 buckets. Do not use periods (.) in your S3 bucket names.

  5. (Optional) Provide additional configuration through the fog_connection hash, which is passed through to the Fog gem.

Fog with NFS

To configure your blobstores to use an NFS-mounted directory, do the following:

  1. Insert the following configuration into your manifest under properties.cc:

    cc:
      buildpacks:
        blobstore_type: fog
        buildpack_directory_key: YOUR-BUILDPACK-DIRECTORY-PREFIX
        fog_connection: &fog_connection
          provider: Local
          local_root: '/var/vcap/data/nfs'
      droplets:
        blobstore_type: fog
        droplet_directory_key: YOUR-DROPLET-DIRECTORY-PREFIX
        fog_connection: *fog_connection
      packages:
        blobstore_type: fog
        app_package_directory_key: YOUR-PACKAGE-DIRECTORY-PREFIX
        fog_connection: *fog_connection
      resource_pool:
        blobstore_type: fog
        resource_directory_key: YOUR-RESOURCE-DIRECTORY-PREFIX
        fog_connection: *fog_connection
    
  2. Replace YOUR-BUILDPACK-DIRECTORY-PREFIX, YOUR-DROPLET-DIRECTORY-PREFIX, YOUR-PACKAGE-DIRECTORY-PREFIX, and YOUR-RESOURCE-DIRECTORY-PREFIX with the names of the directories you will be using. If you do not specify these, Cloud Foundry will supply usable default values. It is important that local_root be set to /var/vcap/data/nfs because the Cloud Controller jobs expect the NFS volume to be mounted at that path.

  3. (Optional) Provide other configuration with the fog_connection hash, which is passed through to the Fog gem.

  4. Configure the nfs_mounter job for the cloud_controller_ng, cloud_controller_worker, and cloud_controller_clock instance groups. The nfs_mounter job specifies how to mount a remote NFS server onto a local directory.

    name: nfs_mounter
      properties:
        nfs_server:
          address: NFS-ENDPOINT
          share: REMOTE-DIR-TO-MOUNT
          share_path: '/var/vcap/data/nfs'
          nfsv4: false
      release: capi
    

Replace the placeholder text in the example above with the values you want to use. Sample values:

  • NFS-ENDPOINT: nfstestserver.service.cf.internal
  • REMOTE-DIR-TO-MOUNT: /var/data/exported

WebDAV

To configure your blobstores to use the WebDAV protocol, perform the steps below:

  1. Ensure your deployment manifest has a single instance of the blobstore job. For a working example, see the example bosh-lite manifest.

  2. Insert the following configuration into your manifest under properties.blobstore and properties.cc:

    blobstore:
      admin_users:
      - password: WEBDAV-BASIC-AUTH-PASSWORD
        username: WEBDAV-BASIC-AUTH-USER
      port: 8080
      secure_link:
        secret: WEBDAV-SECRET
      tls:
        cert: WEBDAV-CERT
        port: 4443
        private_key: WEBDAV-PRIVATE-KEY
        ca_cert: WEBDAV-CA-CERT-BUNDLE
    cc:
      buildpacks:
        blobstore_type: webdav
        buildpack_directory_key: cc-buildpacks
        webdav_config: &webdav_config
          public_endpoint: WEBDAV-PUBLIC-ENDPOINT
          private_endpoint: WEBDAV-PRIVATE-ENDPOINT
          username: WEBDAV_BASIC-AUTH-USER
          password: WEBDAV_BASIC-AUTH-PASSWORD
          ca_cert: WEBDAV-CA-CERT-BUNDLE
      droplets:
        blobstore_type: webdav
        droplet_directory_key: cc-droplets
        webdav_config: *webdav_config
      packages:
        blobstore_type: webdav
        app_package_directory_key: cc-packages
        webdav_config: *webdav_config
      resource_pool:
        blobstore_type: webdav
        resource_directory_key: cc-resources
        webdav_config: *webdav_config
    
  3. Configure your WebDAV blobstores by doing the following:

    • Replace WEBDAV-BASIC-AUTH-USER and WEBDAV-BASIC-AUTH-PASSWORD with Basic AUTH credentials that Cloud Controller can use to communicate with your WebDAV installation.
    • Replace WEBDAV-SECRET with a secret phrase used to sign URLs.
    • Replace WEBDAV-CERT, WEBDAV-PRIVATE-KEY, and WEBDAV-CA-CERT-BUNDLE with proper TLS configuration that are used for the internal blobstore.
    • Replace WEBDAV-PUBLIC-ENDPOINT with the public URL that resolves to your WebDAV installation. For example, https://blobstore.SYSTEM-DOMAIN.example.com.
    • Replace WEBDAV-PRIVATE-ENDPOINT with a routable URL on your internal network. If not set, this defaults to https://blobstore.service.cf.internal:4443.
    • Replace WEBDAV-BASIC-AUTH-USER and WEBDAV-BASIC-AUTH-PASSWORD with Basic AUTH credentials that Cloud Controller can use to communicate with your WebDAV installation.
  4. Create a webdav_config hash to provide further configuration.

View the source for this page in GitHub