Skip to content
/ aio-tvm Public

Token Vending Machine, integrate external cloud services into I/O Runtime

License

Notifications You must be signed in to change notification settings

adobe/aio-tvm

Build Status License Codecov Coverage

Adobe I/O Token Vending Machine (TVM)

This is an implementation of a TVM delivering temporary and restricted tokens to access various cloud services. Users authenticate to the TVM with their Adobe I/O Runtime (a.k.a OpenWhisk) credentials and are only authorized to access their own resources.

A set of Adobe I/O TVM actions are deployed behind the Adobe I/O Gateway at https://firefly-tvm.adobe.io.

Use

  • JavaScript NPM Client: @adobe/aio-lib-core-tvm

  • cURL

    • requirements: valid Adobe I/O Runtime credentials, namespace and auth
    • endpoints: azure/blob/{namespace}, azure/cosmos/{namespace}, aws/s3/{namespace}, azure/presign/{namespace}
curl "https://firefly-tvm.adobe.io/azure/blob/{namespace}" \
  -H "Authorization: Basic {base 64 of auth}" \
  -H "x-Api-Key: firefly-aio-tvm"

Explore

goto API

Deploy your own TVM

Why

You want to share a cloud service that you own (e.g 1 S3 account) with a set of OpenWhisk namespaces and you want to make sure that each namespace has access only to the resources they own (e.g can only see their S3 blobs).

This might be useful for you if:

  • You have multiple Adobe I/O Runtime namespaces and you need them to access a cloud service but you don't want to use the one exposed by Adobe's TVM
  • You are an OpenWhisk provider and want to provide an easy access to an external cloud service (e.g. storage)

Setup

  • npm install -g @adobe/aio-cli
  • npm install

Deployment Config

  • .env:

    # Adobe I/O Runtime deployment credentials
    AIO_RUNTIME_APIVERSION=v1
    AIO_RUNTIME_APIHOST=https://adobeioruntime.net
    AIO_RUNTIME_NAMESPACE=<deployment_ns>
    AIO_RUNTIME_AUTH=<deployment_auth_ns>
    
    # TVM credentials options
    EXPIRATION_DURATION=<token expiration in seconds>
    EXPIRATION_DURATION_STATE_LIB=<State lib expiration token>
    APPROVED_LIST=<comma separated list of namespaces>
    
    # AWS S3 credentials
    AWS_ACCESS_KEY_ID=<key id of IAM user created in AWS>
    AWS_SECRET_ACCESS_KEY=<secret of IAM user created in AWS>
    AWS_REGION=us-east-1
    S3_BUCKET=<MY_BUCKET>
    
    # Azure Storage account credentials
    AZURE_STORAGE_ACCOUNT=<storage account name>
    AZURE_STORAGE_ACCESS_KEY=<storage access key>
    
    # Azure Cosmos Credentials
    AZURE_COSMOS_ACCOUNT=<cosmosdb account name>
    AZURE_COSMOS_MASTER_KEY=<cosmosdb master key>
    AZURE_COSMOS_DATABASE_ID=<cosmosdb database name>
    AZURE_COSMOS_CONTAINER_ID=<cosmosdb database name>
    
    # Adobe I/O API Gateway token validation specific
    DISABLE_ADOBE_IO_API_GW_TOKEN_VALIDATION=<optional, set to true if TVM is not deployed behind the Adobe I/O API Gateway>
    IMS_ENV=<not relevant if DISABLE_ADOBE_IO_API_GW_TOKEN_VALIDATION=true, IMS env for validating the Adobe I/O API Gateway token>
    
    # Optional Params for Monitoring and access control
    AIO_METRICS_URL=<URL where TVM usage metrics will be sent>
    AIO_DENY_LIST_URL=<URL from where to fetch service based deny list>
  • Use the APPROVED_LIST variable to control which namespace can access the TVM and hence who can deploy files to your S3 Bucket.

    • [ ⚠️ NOT RECOMMENDED ⚠️] Use APPROVED_LIST=* to allow access to every OpenWhisk namespace in the same domain.

Setup Azure Blob

  • Create a storage account
  • Retrieve the access key
  • Fill AZURE_STORAGE_* variables in .env

Setup Azure CosmosDB

  • Create a cosmos account
  • Click on + Add Container:
    • specify a database name
    • specify a container name
    • specify a partitionKey: call it /partitionKey => the name is important
    • create the container
  • Configure the newly created container to have Time To live set to On and with a default of 86400 seconds
  • Retrieve the account master key and set AZURE_COSMOS_MASTER_KEY

Setup AWS S3

  • Create a Bucket in S3 that will host the app folders

  • Create an IAM user with the following IAM policy in AWS, replace MY_BUCKET with your bucket name:

      {
        "Version": "2012-10-17",
        "Statement": [
          {
            "Sid": "AllowTokenDelivery",
            "Effect": "Allow",
            "Action": "sts:GetFederationToken",
            "Resource": "*"
          },
          {
            "Sid": "AllowS3",
            "Effect": "Allow",
            "Action": [
                "s3:PutObject",
                "s3:DeleteObject",
                "s3:PutObjectAcl",
                "s3:ListBucket",
            ],
            "Resource": [
              "arn:aws:s3:::MY_BUCKET/*",
              "arn:aws:s3:::MY_BUCKET"
            ]
          }
        ]
      }
  • Configure the .env file, see config

Deploy the TVM endpoints

  • you likely need to undeploy first to refresh the I/O Runtime Api GW

  • aio rt api delete tvm && aio app deploy will deploy all TVM endpoints to the OpenWhisk namespace configured in .env.

Undeploy

  • aio app undeploy to undeploy

Release a new version

  • npm run release will run tests, bump up the version and release tags to github.

Contributing

Contributions are welcomed! Read the Contributing Guide for more information.

Licensing

This project is licensed under the Apache V2 License. See LICENSE for more information.