Skip to content

edgeless-project/runtime-python

Repository files navigation

runtime-python

Run-time environment in Python for EDGELESS.

How to build

The gRPC API specifications must be imported from the core EDGELESS project. This can be done automatically with a script, which will also compile the protobuf data structures and stubs:

scripts/compile-proto.sh

The above step requires protoc and Python gRPC tools, which can be installed with:

pip install grpcio-tools

How to create a Docker container

Make sure you have Docker installed, if not get Docker.

Build the container with:

docker build -t python-function .

Check it with:

docker image ls python-function

How to run

In one shell start the container:

docker run -it --rm --network host python-function

In another launch the EDGELESS node command-line interface emulator:

python3 src/node_cli.py

Examples of commands that you can feed to the emulator:

  • cast recast another-function-alias new-message: cast a message to the function, which instructs the latter to create an asynchronous event on the target another-function-alias with the given message payload
  • stop: invoke the stop() function handler
  • call noret: invoke the call() handler in such a way that a no-return reply is generated
  • cast sync inner-function-state: triggers the function to invoke a state synchronization primitive

When using Docker's host network mode the default values of the command-line arguments of the container and Python scripts should work fine.

If this is not the case, then you can adjust the values to your needs via:

  • direct command-line arguments to the node emulator (try with -h to see the available options)
  • the following environment variables can be passed to Docker using --env ENV_NAME=env_value:
    • ENV LOG_LEVEL
    • ENV PORT
    • ENV MAX_WORKERS

Run without Docker

For development/troubleshooting you can also start directly the function:

python3 src/function.py

Run on Docker without host network

The --network host option makes the function application listens to a port on the host. If this is not possible, then you will have to publish the port:

docker run -it --rm -P python-function

At this point you have to learn what is the public port to which Docker mapped the private port inside the container.

First, get the name of the container with:

docker ps -a -f ancestor=python-function

With some luck, you should be able to retrieve the port mapping of the container with this command:

docker port $(docker ps -a -f ancestor=python-function -q | head -n 1)

which will produce an output like the following:

7101/tcp -> 0.0.0.0:32815
7101/tcp -> [::]:32809

In this example, the private port 7101 for IPv4 was mapped to port 32815.

Then you can run the node emulator with:

python3 src/node_cli.py --function-endpoint 127.0.0.1:32815 --host-endpoint 10.1.1.2:50050

where --function-endpoint specifies the public port mapped by Docker for the function running in the container (32815 in our example) and --host-endpoint must be an end-point with an address that is reachable from the container (run hostname -I for your options) and a free TCP port (50050 in the example).

License

The Repository is licensed under the MIT License. Please refer to LICENSE and CONTRIBUTORS.txt.

Funding

EDGELESS received funding from the European Health and Digital Executive Agency (HADEA) program under Grant Agreement No 101092950.

About

Run-time environment in Python for EDGELESS

Resources

License

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published