mse-cli 1.2

Command Line Interface for Microservice Encryption

Microservice Encryption Cloud Command-Line (MSE)

Overview

Python CLI for Microservice Encryption. See Github repository.

Read the MSE documentation.

Install

$ pip install mse-cli

Usage

You can run MSE CLI to manage microservice deployed on the Cosmian cloud infrastructure using mse cloud or deployed on your own SGX infrastructure using mse home.

MSE Cloud ☁️ is designed to start an MSE application on Cosmian SGX hardware in a fully zero trust environment.

MSE Home 🏕️ is designed to start an MSE application on your own SGX hardware without using all the MSE cloud infrastructure. We explain later how all the subscommands can be chained to deploy your own application.

In that latter usage, two actors are required:

• The code provider (who can also consume the result of the MSE application)
• The SGX operator (who also owns the data to run against the MSE application)

If writing the subcommand part home or cloud every time upsets you, you can set the MSE_DEFAULT_ENV env variable to one of these values and then just omit that word in the future commands. By default, the CLI will target this default environment. For example: mse cloud test turns into mse test

Usage - MSE cloud

$ mse cloud -h

Note: if you declare the env variable MSE_BACKTRACE to the value full, a python stacktrace will be printed in case of errors.

Pre-requisites

First of all sign up or sign in using:

$ mse login

Download the mse-app-examples repository. And go to the helloworld directory.

You can find an example of flask application and a mse.toml configuration file.

Deployment

You can deploy this application as follow:

$ cd helloworld
$ mse deploy --path mse.toml

mse creates <uuid>.toml in $MSE_CONF_PATH/context for each new deployment which contains some context data.

If --path is not provided, mse is expecting a mse.toml in the current directory when using deploy subcommand.

More subcommands

$ mse --help

More parameters

You can use these following env variables:

• MSE_CONF_PATH to use another directory than ~/.config/mse/
• MSE_BASE_URL to use another backend url than https://backend.mse.cosmian.com
• MSE_AUTH0_DOMAIN_NAME to specify another auth0 login url
• MSE_AUTH0_CLIENT_ID to specify another auth0 tenant client id
• MSE_AUTH0_AUDIENCE to specify another tenant audience
• MSE_CONSOLE_URL to specify another console URL
• MSE_PCCS_URL to specify another PCCS URL

Usage - MSE Home

$ mse home -h

Note: if you set the env variable MSE_BACKTRACE=full, a Python stacktrace will be printed in case of errors.

You can find below the use flow step by step.

Scaffold your app

User: the code provider

$ mse home scaffold example

Test your app, your docker and your mse home configuration

User: the code provider

$ mse home localtest --project example/

Create the MSE package with the code and the docker image

User: the code provider

$ mse home package --project example/ \
                   --output workspace/code_provider 

The generated package can now be sent to the sgx operator.

Spawn the MSE docker

User: the SGX operator

$ mse home spawn --host myapp.fr \
                 --port 7777 \
                 --size 4096 \
                 --package workspace/code_provider/package_mse_src_1683276327723953661.tar \
                 --output workspace/sgx_operator/ \
                 app_name

At this moment, evidences have been automatically collected and the microservice is up.

Evidences are essential for the code provider to verify the trustworthiness of the running application.

The file workspace/sgx_operator/evidence.json can now be shared with the other participants.

Check the trustworthiness of the application

User: the code provider

The trustworthiness is established based on multiple information:

• the full code package (tarball)
• the arguments used to spawn the microservice
• evidences captured from the running microservice

Verification of the enclave information:

$ mse home verify --package workspace/code_provider/package_mse_src_1683276327723953661.tar \
                  --evidence output/evidence.json \
                  --output /tmp

If the verification succeeds, you get the RA-TLS certificate (written as a file named ratls.pem) and you can now seal the code key to share it with the SGX operator.

Seal your secrets

User: the code provider

$ mse home seal --secrets example/secrets_to_seal.json \
                --cert /tmp/ratls.pem \
                --output workspace/code_provider/

Finalize the configuration and run the application

User: the SGX operator

$ mse home run --sealed-secrets workspace/code_provider/secrets_to_seal.json.sealed \
               app_name

Test the deployed application

User: the SGX operator

$ mse home test --test workspace/sgx_operator/tests/ \
                --config workspace/sgx_operator/mse.toml \
                app_name

Decrypt the result

User: the code provider

Assume the SGX operator gets a result as follow: curl https://localhost:7788/result --cacert /tmp/ratls.pem > result.enc

Then, the code provider can decrypt the result as follow:

$ mse home decrypt --key key.txt \
                   --output workspace/code_provider/result.plain \
                   result.enc
$ cat workspace/code_provider/result.plain

Manage the MSE docker

User: the SGX operator

You can stop and remove the docker as follow:

$ mse home stop [--remove] <app_name>

You can restart a stopped and not removed docker as follow:

$ mse home restart <app_name>

You can get the MSE docker logs as follow:

$ mse home logs <app_name>

You can get the MSE docker status as follow:

$ mse home status <app_name>

You can get the list of running MSE dockers:

$ mse home list

Development & Test

To work with the development/test environment, you shall edit the following variables with their proper values:

• MSE_CONF_PATH
• MSE_AUTH0_CLIENT_ID
• MSE_AUTH0_DOMAIN_NAME
• MSE_BASE_URL
• MSE_AUTH0_AUDIENCE
• MSE_CONSOLE_URL
• MSE_PCCS_URL

Do the same, if you need to use de staging environment.

Then you first need to login in in order to generate a session. Then run the test.

$ pytest -m 'not home not cloud'
$ mse login
$ export MSE_TEST_DOMAIN_NAME="EDIT"
$ export MSE_TEST_PRIVATE_KEY="EDIT"
$ export MSE_TEST_PUBLIC_KEY="EDIT"
$ pytest -m 'cloud'
$ # The following tests requires a SGX machine on the host
$ export TEST_PCCS_URL="https://pccs.staging.mse.cosmian.com" 
$ export TEST_SIGNER_KEY="/opt/cosmian-internal/cosmian-signer-key.pem"
$ pytest -m 'home'

Documentation

$ cd documentation
$ mkdocs serve

then open your browser on: http://127.0.0.1:8003/

or in PDF:

$ sudo apt install pandoc
$ sudo apt install texlive-xetex texlive-fonts-recommended texlive-fonts-extra
$ cd documentation
$ ./build_pdf.sh

Dockerisation

You can work with mse home without having internet access even to install the CLI by running the CLI docker. Or you can easily run mse cloud by just pulling the CLI docker.

You can build a docker for mse as follow:

$ docker build -t mse . 

Then run it.

In the following, the current directory inside the docker is /mnt/workspace. You can retrieve all generated files in your current host in: /mnt/workspace. Make sure to create it before all at the exact location /mnt/workspace (it will not work if both locations are not aligned).

MSE Cloud

You need to login through the web browser, which is not possible through Docker. You should then run the mse login inside the docker and copy/paste the displayed URL into the host web browser to be logged in inside the docker.

Make sure to mount $HOME/.config/mse into /root/.config/mse to be able to reuse the login credentials or recover your previous deployment information when chaining the commands.

If you need to target another environment, use the docker parameter -e to specify the previous mentionned MSE env variables.

$ # You have to create /mnt/workspace
$ sudo mkdir /mnt/workspace
$ # Log in
$ docker run -v /mnt/workspace:/mnt/workspace \
             -v $HOME/.config/mse:/root/.config/mse \
             -v /var/run/docker.sock:/var/run/docker.sock \
             --network=host \
             mse cloud login
$ # Scaffold an app
$ docker run -v /mnt/workspace:/mnt/workspace \
             -v $HOME/.config/mse:/root/.config/mse \
             -v /var/run/docker.sock:/var/run/docker.sock \
             mse cloud scaffold my_app
$ # Test it
$ docker run -v /mnt/workspace:/mnt/workspace \
             -v $HOME/.config/mse:/root/.config/mse \
             -v /var/run/docker.sock:/var/run/docker.sock\
             --network=host \
             mse cloud localtest --path my_app/mse.toml 
$ # Deploy the app (make sure to use --workspace .)
$ docker run -v /mnt/workspace:/mnt/workspace \
             -v $HOME/.config/mse:/root/.config/mse \
             -v /var/run/docker.sock:/var/run/docker.sock \
             --network=host \
             mse cloud deploy --path my_app/mse.toml \
                              --workspace .
$ # Query it
$ curl https://ae5ed58137d7ec20.dev.cosmian.io/health --cacert /mnt/workspace/ratls.pem
$ # Test it
$ docker run -v /mnt/workspace:/mnt/workspace \
             -v $HOME/.config/mse:/root/.config/mse \
             -v /var/run/docker.sock:/var/run/docker.sock \
             --network=host \
             mse cloud test
$ # Manage the context files
$ docker run -v /mnt/workspace:/mnt/workspace \
             -v $HOME/.config/mse:/root/.config/mse \
             -v /var/run/docker.sock:/var/run/docker.sock \
             mse context --export 631a988f-4001-4593-a023-d316eb57d1f1
$ # Verify the app
$ docker run -v /mnt/workspace:/mnt/workspace \
             -v $HOME/.config/mse:/root/.config/mse \
             -v /var/run/docker.sock:/var/run/docker.sock \
             --network=host \
             mse context --context 631a988f-4001-4593-a023-d316eb57d1f1.toml \
                         --code my_app/mse_src \
                         --workspace . \
                         ae5ed58137d7ec20.dev.cosmian.io 
$ # List your apps
$ docker run -v /mnt/workspace:/mnt/workspace \
             -v $HOME/.config/mse:/root/.config/mse \
             -v /var/run/docker.sock:/var/run/docker.sock \
             mse list 
$ # Get your app status
$ docker run -v /mnt/workspace:/mnt/workspace \
             -v $HOME/.config/mse:/root/.config/mse \
             -v /var/run/docker.sock:/var/run/docker.sock \
             mse status 631a988f-4001-4593-a023-d316eb57d1f1
$ # Get your app logs
$ docker run -v /mnt/workspace:/mnt/workspace \
             -v $HOME/.config/mse:/root/.config/mse \
             -v /var/run/docker.sock:/var/run/docker.sock \
             mse logs 631a988f-4001-4593-a023-d316eb57d1f1
$ # Stop it
$ docker run -v /mnt/workspace:/mnt/workspace \
             -v $HOME/.config/mse:/root/.config/mse \
             -v /var/run/docker.sock:/var/run/docker.sock \
             mse stop 631a988f-4001-4593-a023-d316eb57d1f1

MSE Home

$ # You have to create /mnt/workspace
$ sudo mkdir /mnt/workspace
$ # Scaffold a project
$ docker run -v /mnt/workspace:/mnt/workspace \
             -v /var/run/docker.sock:/var/run/docker.sock \
            mse home scaffold example
$ # Test it
$ docker run -v /mnt/workspace:/mnt/workspace \
             -v /var/run/docker.sock:/var/run/docker.sock \
             --network=host \
             mse home localtest --project example
$ # Package it
$ docker run -v /mnt/workspace:/mnt/workspace \
             -v /var/run/docker.sock:/var/run/docker.sock \
             mse home package --project example \
                              --output .
$ # Spawn it
$ docker run -v /mnt/workspace:/mnt/workspace \
             -v /var/run/docker.sock:/var/run/docker.sock \
             --network=host \
             mse home spawn --host localhost \
                            --port 7779 \
                            --days 365 \
                            --signer-key /opt/cosmian-internal/cosmian-signer-key.pem \
                            --size 32768 \
                            --pccs https://pccs.staging.mse.cosmian.com \
                            --package package_example_1688025211482089576.tar \
                            --output . \
                            test_docker
$ # Verify the evidences
$ docker run -v /mnt/workspace:/mnt/workspace \
             -v /var/run/docker.sock:/var/run/docker.sock \
             mse home verify --package package_example_1688025211482089576.tar \
                             --evidence evidence.json \
                             --output .
$ # Seal the secrets
$ docker run -v /mnt/workspace:/mnt/workspace \
             -v /var/run/docker.sock:/var/run/docker.sock \
             mse home seal --secrets example/secrets_to_seal.json \
                           --cert ratls.pem \
                           --output  .
$ # Complete the configuration
$ docker run -v /mnt/workspace:/mnt/workspace \
             -v /var/run/docker.sock:/var/run/docker.sock \
            --network=host \
            mse home run --sealed-secrets secrets_to_seal.json.sealed test_docker
$ # Query it
$ curl https://localhost:7779 --cacert /mnt/workspace/ratls.pem 
$ # Test it
$ docker run -v /mnt/workspace:/mnt/workspace \
             -v /var/run/docker.sock:/var/run/docker.sock \
             --network=host \
             mse home test --test example/tests \
                           --config mse.toml test_docker
$ # Remove it
$ docker run -v /var/run/docker.sock:/var/run/docker.sock \
             mse home stop --remove test_docker

Use case

Let's assume your microservice have to interface with a frontend. The main issue you will face up is that the RA-TLS certificate is not allowed in your web browser, mainly because it's a self signed cert. Also the RA-TLS extension is not checked by your webbrowser, yet any query to the webservice must verify the RA-TLS extension of the certificate: the security is based on this verification.

Therefore, the frontend can't interact straightaway with your webservice through the web browser. You need to develop a intermediate backend acting like a proxy. Or a simpler way could be to use the following nginx configuration:

server {
    listen       8080 default_server;
    listen       [::]:8080 default_server;
    server_name  _;

    location / {
         proxy_pass	https://9cc36ebf3c351eba.dev.cosmian.io;
	     # We need the next two lines because the RA-TLS certificate
         proxy_ssl_trusted_certificate /tmp/tmp4b0174yn/ratls.pem;
         proxy_ssl_verify on;
	     # We need the next two lines to allow sni routing by the haproxy
         proxy_ssl_name 9cc36ebf3c351eba.dev.cosmian.io;
         proxy_ssl_server_name on;
    }
}