Configuration

Advanced configuration options for the KFP Operator platform

Platform Configuration

This section provides comprehensive guidance for configuring the KFP Operator platform for production use. As a platform engineer, you’ll learn how to customize the operator behavior, set up providers, and optimize for your specific environment.

📋 Configuration Overview

The KFP Operator configuration consists of several layers:

1. Operator Configuration

Core operator settings and behavior

  • Operator Configuration - Main operator settings
  • Resource allocation and scaling
  • Logging and monitoring configuration
  • Webhook and admission control settings

2. Provider Configuration

ML orchestration platform integrations

3. Security Configuration

RBAC, policies, and compliance settings

  • Security Setup - Enterprise security patterns
  • Service account configuration
  • Network policies and isolation
  • Secret and credential management

4. Monitoring Configuration

Observability and alerting setup

  • Monitoring Setup - Comprehensive observability
  • Metrics collection and export
  • Log aggregation and analysis
  • Alerting rules and notifications

Quick Configuration Guide

Basic Production Setup

1. Operator Configuration

# values.yaml - Production configuration
manager:
  replicas: 2  # High availability
  resources:
    requests:
      cpu: 200m
      memory: 256Mi
    limits:
      cpu: 1000m
      memory: 1Gi
  
  # Enable monitoring
  monitoring:
    create: true
    serviceMonitor:
      create: true

# Enable event system
statusFeedback:
  enabled: true

# Production logging
logging:
  verbosity: 1

2. Provider Setup

# kubeflow-provider.yaml
apiVersion: pipelines.kubeflow.org/v1beta1
kind: Provider
metadata:
  name: production-kfp
  namespace: kfp-operator-system
spec:
  type: kfp
  kfp:
    restKfpApiUrl: "https://kubeflow.company.com/pipeline"
    uiUrl: "https://kubeflow.company.com/"

3. RBAC Configuration

# Enable RBAC with custom roles
manager:
  rbac:
    create: true
    additionalClusterRoles:
      - name: kfp-operator-viewer
        rules:
          - apiGroups: ["pipelines.kubeflow.org"]
            resources: ["pipelines", "runs"]
            verbs: ["get", "list", "watch"]

Environment-Specific Configurations

Development Environment

# dev-values.yaml
manager:
  replicas: 1
  resources:
    requests:
      cpu: 100m
      memory: 128Mi
    limits:
      cpu: 500m
      memory: 512Mi

logging:
  verbosity: 2  # Debug logging

# Relaxed timeouts for development
argo:
  stepTimeoutSeconds:
    default: 3600  # 1 hour

Staging Environment

# staging-values.yaml
manager:
  replicas: 1
  resources:
    requests:
      cpu: 150m
      memory: 192Mi
    limits:
      cpu: 750m
      memory: 768Mi

# Production-like settings with shorter retention
argo:
  ttlStrategy:
    secondsAfterCompletion: 1800  # 30 minutes

Production Environment

# prod-values.yaml
manager:
  replicas: 3  # High availability
  resources:
    requests:
      cpu: 300m
      memory: 512Mi
    limits:
      cpu: 1500m
      memory: 2Gi

# Production monitoring
monitoring:
  create: true
  serviceMonitor:
    create: true
    interval: 30s

# Strict security
rbac:
  create: true
webhookCertificates:
  provider: cert-manager

🆘 Configuration Troubleshooting

Common Configuration Issues

Invalid Helm Values

# Debug Helm template rendering
helm template kfp-operator kfp-operator/kfp-operator \
  --values values.yaml \
  --debug \
  --output-dir ./debug-output

# Check for YAML syntax errors
yamllint values.yaml

Provider Connection Issues

# Test provider connectivity
kubectl run debug-pod --image=curlimages/curl --rm -it -- \
  curl -v http://kubeflow-provider-url/api/v1/healthz

# Check DNS resolution
kubectl run debug-pod --image=busybox --rm -it -- \
  nslookup kubeflow-provider-hostname

Getting Help

For configuration issues:

  1. Check Troubleshooting Guide for platform issues
  2. Review Architecture Documentation for system understanding
  3. Consult Security Guide for RBAC and security issues
  4. Ask in GitHub Discussions for community help

Ready to configure your platform? Start with the Operator Configuration guide for core settings, then move on to Provider Setup for ML platform integration.

Operator Configuration

Advanced configuration options for the KFP Operator core components

Providers