ai-runner-app

Currently the ai-runner app must be run as a user having admin level kubernetes access to create and delete namepaces, PVCs and nodeport level services. Moreover, there must be a default storage provider which provides volumes with RWX access mode, so that all nodes of the same pipeline can read and write.

1. Install Python requirements

AI_RUNNER_VENV=~/.grpc
virtualenv -p /usr/bin/python3 $AI_RUNNER_VENV
source $AI_RUNNER_VENV/bin/activate
pip install -r requirements.txt 

2. Download a solution and adapt pathSolutionZip in exampleObjectModels.py accordingly.

3. Run python exampleObjectModels.py to start at least one pipeline. You can see the pipeline running with kubectl get ns. The namespace with the pipeline name in the beginning will contain the running pipeline.

4. Run python app.py to start the playground app, which will be available at 127.0.0.1:5000

Cleanup

To cleanup the namespaces, you can run

python -m scripts.cleanup_all_solutions

Note that you can choose between namespaces by considering solution_folders or namespaces considering a regex which should match all namespace names created by the playground.