Skip to content

Latest commit

 

History

History
145 lines (94 loc) · 8.1 KB

README.md

File metadata and controls

145 lines (94 loc) · 8.1 KB
Raw

API Endpoints

  • public keys are formatted with the Protobuf scheme found in proto/src/request_protos/pubkeys.proto at this URL:

Limitations

This licensing API is currently limited to the following license types:

  • Trial
  • Subscription
  • Perpetual

Offline perpetual license limitiations

There isn't a way to remove/deactivate machines on a license (yet), but when there is a way to do that, there will be a conflict with Offline licenses. Offline licenses are licenses where a client computer will have an activated license indefinitely without needing to contact the server periodically to renew the license's expiration. That means that if someone had an offline computer with an offline license activated, then they could "deactivate" this offline computer, and since the offline computer won't talk to the server any more, the offline computer's license will remain activated forever. This could also happen if someone set up a firewall that blocked communication to the licensing service on the client's network or local machine.

There are a few ways to address this issue:

  • use a physical device for offline licenses
    • we aren't like other competitors. we don't want to put the "plug in" in "plugin"
  • disable deactivation for offline computers
    • this might be an inconvenience, and stores would have to inform users that this is a policy for the software licensing
  • disallow offline licenses
    • it is possible to not have any Offline licenses for a product. It is an option in the create_product API method

There was a plan to allow offline activations by having the user type in their license code followed by -offline-XXXX, but it has since been changed in #43. To activate an offline computer, we now require some server-side code to sign a license_activation_refactor request with the store's private key.

Subscription license limitations

Currently, subscription licenses can only have the base amount of machines using a license. This is because it is difficult to determine whether a subscription license create_license request is meant to purchase a new license, or if it is meant to extend an existing license. This can probably be broken up into 2 separate API methods: one for purchasing licenses, and one for extending subscriptions. However... IMO, if a person was willing to pay for a subscription for some amount of time and stopped paying... you have already earned as much as you might earn from this customer... why not let them keep their license, giving subscription license customers a perpetual license instead of a subscription license?

Building

There are at least two ways to build this for the Graviton 2 processor using cross or cargo-lambda. cargo-lambda compiles a bit faster, but the files are a bit larger and the binaries run a little bit slower. cargo-lambda build scripts are saved as build-cargo-lambda.sh and the cross build scripts are saved as build.sh. You can also pass an argument that specifies the features to use. Available features are currently only zeroize and logging. Here's an example of how to call them:

./build.sh
# or
./build.sh "zeroize"
# or
./build.sh "zeroize,logging"
# or
./build-cargo-lambda.sh "zeroize,logging"

Note: build-cargo-lambda.sh requires provided.al2023 for the Lambda function's runtime, not just provided.al2. The runtime is set to provided.al2023 in the deployment scripts.

To build these API methods, you will need to install a few packages if you don't already have them installed:

  • Install rust on Ubuntu:
curl --proto '=https' --tlsv1.2 -sSf https://sh.rustup.rs | sh
  • Install openssl on Ubuntu:
sudo apt update
sudo apt install build-essential pkg-config libssl-dev
  • Install cargo lambda (optional, and its built binaries do not work at the moment)
rustup update
cargo install cargo-lambda

  • Install Docker Desktop on your Host OS and enable its use in WSL if you are using WSL (highly recommended if you're on Windows). Or you can install zig using Homebrew on Linux/WSL.

  • Install cross

cargo install cross
  • Install aarch64 build target:
rustup target add aarch64-unknown-linux-musl
  • Install aws-cli:
curl "https://awscli.amazonaws.com/awscli-exe-linux-x86_64.zip" -o "awscliv2.zip"
unzip awscliv2.zip
sudo ./aws/install

Setting up local environment variables:

Several environment variables need to be set for this to work:

  1. KEY_MANAGER_PRIVATE_KEY - needs to be a minimum length of the key manager's hash function's output, and a maximum of the hash function's internal buffer. This is currently using sha3_512 since it is faster on aarch64 with inline assembly. So for this, ideally the length should be between 64-72 bytes (inclusive).

  2. STORE_TABLE_SALT - a salt for the stores table

  3. LICENSE_TABLE_SALT - a slat for the licenses table

Running the code on AWS requires setting each Lambda's environment variables.

Getting this to work

There are a few things that you would need to do to get this to work, besides building it.

Here is a non-comprehensive list of what you would need to do to get the refactored version of the code to work:

  1. make an AWS account if you do not have one

  2. create some DynamoDB tables with table names, primary keys, and global secondary keys specified in utils/src/tables/, or change the names in those files to use different table names. You can generate some table names with cargo test --features local -- --nocapture. The long and random table names provide a little entropy when generating resource encryption keys when encrypting some of the data in the tables. It isn't completely necessary to have long table names, but AWS does not charge by the character in the table names. Yes, AWS supposedly encrypts the tables at rest, but why not add an extra layer of encryption? And, believe it or not, the encrypted protobuf messages can actually be smaller in size than putting it in plaintext NoSQL due to the potentially long keys since Protobuf keys are just binary numbers. The downside is that analytics tools such as AWS Athena likely will not be able to analyze any Protobuf/encrypted data.

  3. Create an IAM Role that can read and write to these tables. This is best done by first creating an IAM Policy with DynamoDB permissions.

Permissions for the DynamoDB tables:

  • BatchGetItem
  • BatchWriteItem
  • DeleteItem - only required for Licenses table
  • GetItem
  • Query - only required for Licenses table
  • PutItem
  • UpdateItem

The Query permission also needs to be added for the secondary index of the Licenses table.

Permissions for the Public Keys S3 bucket:

  • PutObject

The public keys bucket needs to be specified in the environment variables for the publish_rotating_keys Lambda Function.

  1. Create an IAM user with permissions for Lambda>Create Function and IAM>Pass Role, then make an access key for this user to sign into the AWS CLI with. Consider using aws configure sso instead, but it isn't very intuitive.

  2. Deploy the lambda functions. First, call create_deployment_scripts, specifying the IAM Role that can access the tables. Then call build.sh and deploy.sh to deploy the functions to the cloud. If anything changes, you need to call update_func.sh to update the functions.

  3. Navigate to API Gateway and create an HTTP API or REST API. Do some research on the two, but you'll probably want a REST API. The differences are explained here

  4. Add the lambda functions to this API, and ensure that these API endpoints are accessible from the public internet.

  5. Optionally, configure AWS WAF to restrict the (ab)usage of the API. You don't want to get DOS-ed.

  6. Optionally, take AWS Lambda Power Tuning for a spin to potentially increase the speed and lower the cost of the lambda functions.