Skip to content

Installation

This guide covers how to deploy the Predictive Pod Autoscaler (PPA) using the Helm chart from our OCI registry.

1. Configure Storage

PPA requires S3-compatible storage for model artifacts. Create a values.yaml file with your storage credentials:

s3:
  endpoint: s3.your-domain.com
  bucket: ppa-models
  accessKey: YOUR_ACCESS_KEY
  secretKey: YOUR_SECRET_KEY
  insecure: false

2. Deploy PPA

Install PPA in your cluster using the OCI Helm chart:

kubectl create namespace ppa
helm install ppa oci://ghcr.io/mehdirtal/predictive-pod-autoscaler/ppa --version 0.1.9 -n ppa -f values.yaml
Quick Install (Optional Dependencies)

If you don't have the prerequisites installed, the PPA chart can install them for you: 

helm install ppa oci://ghcr.io/mehdirtal/predictive-pod-autoscaler/ppa --version 0.1.9 \
  --set keda.enabled=true \
  --set argo-workflows.enabled=true \
  --set seaweedfs.enabled=true

3. Verify Deployment

Confirm that the components are running:

kubectl get pods -l app.kubernetes.io/name=ppa-scaler -n ppa
kubectl get resourcegraphdefinitions

Lifecycle Management

Upgrading

To upgrade to a newer version: 

helm upgrade ppa oci://ghcr.io/mehdirtal/predictive-pod-autoscaler/ppa --version 0.1.9 -n ppa -f values.yaml

Uninstallation

To remove PPA from your cluster: 

helm uninstall ppa -n ppa