mlops-toolkit

Development guide

Prerequisite

1. Python 3.10: Ensure that Python 3.10 is installed on your system before running the application.
2. kubectl installed and configured for your cluster

Set up the mlops-data-fetcher.

cd mlops-toolkit
make proto

Download client certificates for storage from the k9s cluster you want to connect to.

1. You should have configured your kubectl to connect to the k8s cluster.
2. Make sure you have selected the correct cluster

kubectl config set current-context <name-of-oyour-cluster>

3. Run the download_certs.sh script.

./scripts/download_certs.sh [MLOPS_PREFIX] [MLOPS_K8_NAMESPACE]

ex: For xy-test platform instances

./scripts/download_certs.sh mlops default 

Verify that after running the script, three files named ca, certificate, and key have been downloaded into the certs directory from the specified cluster.

Set environment variables.

1. You have to start kubectl port-forwarding for storage.

kubectl port-forward service/<mlops-storage-service-name> 9999:9999

2. Then set the following envs for the app.

export STORAGE_URL=<storage-url-with-port>

ex:

export STORAGE_URL=localhost:9999

Run the app

make run