Skip to content

installation.md

Latest commit

 

History

History
146 lines (104 loc) · 3.25 KB

installation.md

File metadata and controls

146 lines (104 loc) · 3.25 KB

Installation Guide

This guide covers installing Workload-Variant-Autoscaler (WVA) on your Kubernetes cluster.

Prerequisites

  • Kubernetes v1.32.0 or later
  • Helm 3.x
  • kubectl configured to access your cluster
  • Cluster admin privileges

Installation Methods

Option 1: Helm Installation

See the Helm Installation for detailed instructions.

Verify the installation:

kubectl get pods -n workload-variant-autoscaler-system

Option 2: Kustomize Installation

Using kustomize for more control:

# Install CRDs
make install

# Deploy the controller
make deploy IMG=quay.io/llm-d/llm-d-workload-variant-autoscaler:latest

Option 3: Local Development (Kind Emulator):

See the Kind Emulator for detailed instructions.

Configuration

Helm Values

Key configuration options:

# custom-values.yaml
image:
  repository: quay.io/llm-d/llm-d-workload-variant-autoscaler
  tag: latest
  pullPolicy: IfNotPresent

resources:
  limits:
    cpu: 500m
    memory: 512Mi
  requests:
    cpu: 100m
    memory: 128Mi

# Enable Prometheus monitoring
prometheus:
  enabled: true
  servicemonitor:
    enabled: true

# Optional: Multi-controller isolation
# Set a unique identifier for this controller instance
# Useful for parallel testing or multi-tenant environments
# See docs/user-guide/multi-controller-isolation.md
wva:
  controllerInstance: ""  # Leave empty for single controller

ConfigMaps

WVA uses ConfigMaps for cluster configuration:

  • Service Classes: SLO definitions for different service tiers

See Configuration Guide for details.

Integrating with HPA/KEDA

WVA can work with existing autoscalers:

For HPA integration: See HPA Integration Guide

For KEDA integration: See KEDA Integration Guide

Verifying Installation

  1. Check controller is running:

    kubectl get deployment -n workload-variant-autoscaler-system

  2. Verify CRDs are installed:

    kubectl get crd variantautoscalings.llmd.ai

  3. Check controller logs:

    kubectl logs -n workload-variant-autoscaler-system \
  deployment/workload-variant-autoscaler-controller-manager

Uninstallation

Helm:

helm uninstall workload-variant-autoscaler -n workload-variant-autoscaler-system

Kustomize:

make undeploy
make uninstall  # Remove CRDs

Troubleshooting

Common Issues

Controller not starting:

  • Check if CRDs are installed: kubectl get crd
  • Verify RBAC permissions
  • Check controller logs for errors

Metrics not appearing:

  • Ensure Prometheus ServiceMonitor is created
  • Verify Prometheus has proper RBAC to scrape metrics
  • Check network policies aren't blocking metrics endpoint

See Also:

Next Steps