Spinning up a cluster in easy, whether you want to try ML Studio locally or use it in the cloud.
You need to have Google Cloud SDK for this step, follow the instructions to install it here.
export cluster_name=mlstudio-clusterexport cluster_zone=us-central1-agcloud container clusters create $cluster_name \--machine-type=n1-standard-4 \--num-nodes 1 \--enable-autoscaling --min-nodes 0 --max-nodes 6 \--zone $cluster_zone# Chnage kubectl current contextgcloud container clusters get-credentials $cluster_name --zone $cluster_zone
For cost savings you can use --preemptible
nodes.
They offer the same machine types and options as regular compute instances and last for up to 24 hours.
export cluster_name=mlstudio-clusterexport cluster_zone=us-central1-agcloud container clusters create $cluster_name \--machine-type=n1-standard-4 \--preemptible \--num-nodes 1 \--enable-autoscaling --min-nodes 0 --max-nodes 6 \--zone $cluster_zone# Chnage kubectl current contextgcloud container clusters get-credentials $cluster_name --zone $cluster_zone
You can instead create a GPU accelerated cluster by appending --accelerator type=nvidia-tesla-t4,count=1
to the previous command. And then creating a DaemonSet
to instal Nvidia drivers.
kubectl apply -f https://raw.githubusercontent.com/GoogleCloudPlatform/container-engine-accelerators/master/nvidia-driver-installer/cos/daemonset-preloaded.yaml
So the final commands to create a GPU accelerated cluster would look like:
export cluster_name=mlstudio-clusterexport cluster_zone=us-central1-agcloud container clusters create $cluster_name \--machine-type=n1-standard-4 \--accelerator type=nvidia-tesla-t4,count=1 \--num-nodes 1 \--enable-autoscaling --min-nodes 0 --max-nodes 6 \--zone $cluster_zone# Install NVIDIA GPU device driverskubectl apply -f https://raw.githubusercontent.com/GoogleCloudPlatform/container-engine-accelerators/master/nvidia-driver-installer/cos/daemonset-preloaded.yaml# Chnage kubectl current contextgcloud container clusters get-credentials $cluster_name --zone $cluster_zone
minikube start --cpus 5 --memory 10096
Download Docker Desktop application and follow the instructions on enabling Kuberentes.
You need a minimum of:
5 CPUs
8 Gb of RAM
15 Gb of fee desk space