For this workshop you will need Docker and Docker Compose running on your machine. (on mac os and Windows docker-compose is installed with Docker by default)
To build the application Docker image and start the application container, as well as Prometheus and Grafana together, run the following command (from the root of this repo):
docker compose upIf you see errors it may be because you still have the previous version of the application running and therefore might be using the same port as you are now trying to access with Docker. Or the ports interfere with local installations. A local Grafana installation probably runs on port 3000 as a service.
You should then be able to access the Prometheus dashboard on http://localhost:9090 and Grafana on http://localhost:3000.
Create an mlops-workshop conda environment in the cloned project directory by executing:
$ conda env create -f environment.yml
For Apple with an M1 processor install the the appropriate Miniconda version. You can create the mlops-workshop conda environment executing:
$ conda env create -f environment-m1.yml
Either way you need to activate the environment after installation:
$ conda activate mlops-workshop
Monitoring is based on https://github.com/evidentlyai/evidently/tree/main/examples/integrations/grafana_monitoring_service and code is copied from there. Start monitoring as described there.