Skip to content

Latest commit

 

History

History
66 lines (49 loc) · 2.19 KB

installation.md

File metadata and controls

66 lines (49 loc) · 2.19 KB
Raw

Installation

Requirements

The project comes with a quick and easy to build Docker image, which can be built and used in less than a minute. Although, for those who does not want to use the image and run the module locally, the requirements are:

  • requirements.txt
  • python3-pytest (only if testing)
  • BCC (read following lines)

Despite existing a python3-bpfcc packet from the Ubuntu package manager apt, it is not the latest release of such tool. Thus, many of the advanced operations with eBPF maps are not possible, and the framework would fail starting due to an import error. I strongly suggest downloading the latest code from their GitHub page and compile it as follows.

Install

The usage of the Docker images is strongly recommended, to avoid installing the entire BCC dependency locally.

Local

# Installing BCC
git clone https://github.com/iovisor/bcc.git
cd bcc
sudo /usr/lib/pbuilder/pbuilder-satisfydepends
sudo ./scripts/build-deb.sh release
sudo dpkg -i *.deb
cd .. 

# Installing other dependencies
sudo apt install python3-pytest
sudo pip3 install -r requirements.txt
sudo python3 -m dechainy

If you want to install DeChainy as a Python package, after satisfying the BCC dependency, you can use the setup.py script:

sudo python3 setup.py install

From now on, you can reference to this framework as you would do for any other Python package, like import numpy.

Docker

docker build -f Dockerfile -t s41m0n/dechainy:latest .
docker run --rm --privileged --network host \
    -v /lib/modules:/lib/modules:ro \
    -v /etc/localtime:/etc/localtime:ro \
    -v /usr/src:/usr/src:ro \
    -v $(pwd)/dechainy:/app/:ro \ # you can mount at runtime the new code you develop, instead of rebuilding it
    -v $(pwd)/startup.json:/app/startup.json:ro \ # or you can just mount only the startup configuration
    s41m0n/dechainy:latest

If you are willing to use TensorFlow or Keras (using only CPU support by now), the main Dockerfile accepts an additional parameter to include such additional packages:

docker build --build-arg DEFAULT_BUILDTYPE=ml -t s41m0n/dechainy:ml .