Skip to content

Latest commit

 

History

History
96 lines (66 loc) · 3.12 KB

README.md

File metadata and controls

96 lines (66 loc) · 3.12 KB
Raw

Building Kind cluster with everything installed

Pre-requirements

As an alternative, you can execute the following manual installation instructions:

Create cluster

Run the following command to create the cluster:

cd /tmp
git clone https://github.com/IBM/data-prep-kit.git
cd data-prep-kit
ROOT_DIR=$PWD/kind/
kind create cluster --name dataprep --config ${ROOT_DIR}/hack/kind-cluster-config.yaml

Note that by default this will create a kind cluster with 2 worker nodes. If you would like a different amount of node, modify cluster configuration

Install KFP

Install Kubeflow Pipelines and wait for it to be ready:

cd $ROOT_DIR/hack/tools/ && PIPELINE_VERSION=1.8.5 ./install_kubeflow.sh deploy && cd -
kubectl wait --for=condition=ready --all pod -n kubeflow --timeout=300s

Install KubeRay

Install Kuberay:

cd $ROOT_DIR/hack/tools && KUBERAY_APISERVER=1.1.0 KUBERAY_OPERATOR=1.0.0 ./install_kuberay.sh deploy && cd -
kubectl wait --for=condition=ready --all pod -n kuberay --timeout=300s

Install NGNIX

To access the API server and Kubeflow pipeline UI externally, we make use NGINX ingress.

Install Ingress NGNIX for KFP, RAY and MinIO and wait for it to be ready:

${ROOT_DIR}/hack/tools/install_nginx.sh deploy
kubectl wait --namespace ingress-nginx \
          --for=condition=ready pod \
          --selector=app.kubernetes.io/component=controller \
          --timeout=90s

To deploy the ingress for ray apiserver, kfp and Minio execute the following:

kubectl apply -f $ROOT_DIR/hack/ray_api_server_ingress.yaml
kubectl apply -f $ROOT_DIR/hack/kfp_ingress.yaml
kubectl apply -f $ROOT_DIR/hack/minio_ingress.yaml

Open the Kubeflow Pipelines UI at http://localhost:8080/

Working with a MinIO server instead of S3 storage

You can work with a real S3 storage, but for testing you can use the Mino server which is deployed as part of the KFP installation. You can access the Minio dashboard at http://localhost:8090/

Create a secret

The MinIO service, deployed as a part of KFP, uses a username (minio) as an access_key/password (minio123) as the secret key. A secret needs to be created for accessing MinIO using the following command:

kubectl apply -f $ROOT_DIR/hack/s3_secret.yaml

Copy test data

Populating Minio server with test data can be done using mc. Use the following command:

$ROOT_DIR/hack/populate_minio.sh

This file creates an mc alias, creates the test bucket and copies the local test data into MinIO. If you need to load additional data, please load it using additional mc commands, similar to the ones being used by populate_minio.sh

Cleanup

See Clean up the cluster