values.yaml

router:
  modelServers:
    # matchLabels: # REQUIRED
    #   app: vllm-qwen3-32b

# Gateway provider selection.
# This section configures the gateway implementation only; it is not a generic infrastructure provider setting.
# Options: ["gke", "istio", "none"]
provider:
  name: none

  # Istio-specific configuration.
  # This block is only used if name is "istio".
  istio:
    destinationRule:
      # Provide a way to override the default calculated host
      host: ""
      # Optional: Enables customization of the traffic policy
      trafficPolicy: {}
        # connectionPool:
        #   http:
        #     maxRequestsPerConnection: 256000

  # GKE-specific configuration.
  # This block is only used if name is "gke".
  gke:
    preferredBackends:
      enabled: false
      # Number of primary active pod instances pinned to the PREFERRED load balancing tier.
      # Setting >1 scales concurrent active routing.
      preferredReplicas: 1
      # Number of standby pod instances pinned to the DEFAULT load balancing tier.
      # Setting >1 scales warm standby failover capacity.
      defaultReplicas: 1
      # Benchmark capacity and threshold settings for PREFERRED load balancing tier.
      # When incoming traffic exceeds capacityScalerPercent of maxRatePerEndpoint, excess traffic spills over to DEFAULT standby pods.
      balancingMode: RATE
      maxRatePerEndpoint: 100
      capacityScalerPercent: 100

# httpRoute section is used to deploy an HTTPRoute resource as part of the gateway chart.
httpRoute:
  create: false # a flag to indicate whether to create the httproute as part of the chart or not.
  inferenceGatewayName: inference-gateway
  inferenceGatewayNamespace: ""
  # HTTPRoute request timeout. Defaults to "300s". Unused when provider.name is gke.
  requestTimeout: "300s"
  # Example header matches:
  # headerMatches:
  #  x-my-custom-route-header: route-name