Configuration Settings

Configure and customize your Bud Runtime deployment

Configuration Overview

Bud Runtime uses a hierarchical configuration system with multiple levels:

1. Default values
2. Helm values
3. Environment variables
4. Runtime configuration

Helm Configuration

Core Components

Configure the main components in your values.yaml:

# Gateway configuration
budgateway:
  replicaCount: 2
  service:
    type: LoadBalancer
    port: 8080
  autoscaling:
    enabled: true
    minReplicas: 2
    maxReplicas: 10

# Model server configuration
budserve:
  replicaCount: 3
  resources:
    limits:
      nvidia.com/gpu: 1
      memory: 16Gi
    requests:
      nvidia.com/gpu: 1
      memory: 8Gi

# Dashboard configuration
budui:
  enabled: true
  ingress:
    enabled: true
    hostname: dashboard.example.com

Model Configuration

Define model deployments:

models:
  - name: llama2-7b
    replicas: 2
    gpu: true
    memory: 16Gi
  - name: stable-diffusion-xl
    replicas: 1
    gpu: true
    memory: 24Gi

Environment Variables

Key environment variables for runtime configuration:

# API Configuration
BUD_API_KEY=your-api-key
BUD_API_BASE_URL=https://api.bud.studio

# Model Cache
BUD_MODEL_CACHE_PATH=/models
BUD_CACHE_SIZE_GB=100

# Logging
BUD_LOG_LEVEL=info
BUD_LOG_FORMAT=json

Runtime Configuration

API Rate Limiting

rateLimit:
  enabled: true
  requests:
    perMinute: 60
    perHour: 1000
  burst: 100

Model Loading

modelLoading:
  preload: true
  lazyLoad: false
  timeout: 300s
  retries: 3

Security Settings

Authentication

auth:
  enabled: true
  type: jwt
  issuer: https://auth.example.com
  audience: bud-runtime

TLS Configuration

tls:
  enabled: true
  certManager:
    enabled: true
    issuer: letsencrypt-prod

Monitoring and Observability

Metrics

metrics:
  enabled: true
  serviceMonitor:
    enabled: true
    interval: 30s

Logging

logging:
  level: info
  format: json
  outputs:
    - stdout
    - file: /var/log/bud/runtime.log

Advanced Settings

GPU Scheduling

gpu:
  scheduling:
    strategy: binpack
    nodeSelector:
      gpu-type: nvidia-a100

Storage Configuration

storage:
  modelCache:
    size: 500Gi
    storageClass: fast-ssd
  outputs:
    size: 100Gi
    storageClass: standard

For deployment-specific configurations, refer to the Installation Guide.