FAQ¶

Installation¶

Install Gateway API¶

Via kpanda Helm Template¶

gateway-api-helm-charts

install-gateway-api

Via Online Installation¶

GATEWAY_API_VERSION=v1.5.0
kubectl apply --server-side -f https://github.com/kubernetes-sigs/gateway-api/releases/download/${GATEWAY_API_VERSION}/standard-install.yaml

Install agent-gateway¶

AGW_VERSION=v1.0.0-alpha.4
helm upgrade -i --create-namespace --namespace agentgateway-system --version $AGW_VERSION agentgateway-crds oci://cr.agentgateway.dev/charts/agentgateway-crds
helm upgrade -i --namespace agentgateway-system --version $AGW_VERSION agentgateway oci://cr.agentgateway.dev/charts/agentgateway --set inferenceExtension.enabled=true

Install LWS¶

VERSION=v0.8.0
helm install lws https://github.com/kubernetes-sigs/lws/releases/download/$VERSION/lws-chart-$VERSION.tgz --namespace lws-system --create-namespace

inferencepools CRD Conflict During Installation¶

You can skip the inferencepools CRD installation by adding the --skip-crds parameter:

export INFERX_CHART_VERSION=0.1.0-xxx
helm -n public upgrade --install qwen3-06b inferx/inferx --version $INFERX_CHART_VERSION --skip-crds -f manifests/examples/inference-scheduling/values-single-vgpu.yaml

Service Unavailable¶

When Using HAMI & PD Disaggregation, Decode Pod Cannot Be Scheduled Due to SidecarContainers Feature Not Enabled¶

The error message is as follows:

Warning  FailedScheduling  2m56s                hami-scheduler  0/1 nodes are available: 1 Pod has a restartable init container and the SidecarContainers feature is disabled. preemption: 0/1 nodes are available: 1 Preemption is not helpful for scheduling..

The kube-scheduler image version in the current cluster's nvidia-vgpu-hami-scheduler Pod is too low and does not support the SidecarContainers feature. Upgrade the image version to 1.29 or higher.

HTTPRoute Error: Group `inference.networking.k8s.io` or `inference.networking.x-k8s.io` Not Supported¶