Skip to navigation

This page covers Helm values that apply to any Studio Voice NIM service chart deployment. The same keys appear under nvidia-studio-voice-h4m-service.<key> when the NIM is deployed as part of the end-to-end demo chart.


Global Overrides

Both the NIM service chart and the end-to-end sample (umbrella) chart expose a global section. When a global value is non-empty, it overrides the corresponding per-component key for all subcharts (sender, receiver, and NIM service). Leave a value as "" to use each component’s own setting instead.

ConfigurationHelm KeyOverridesComment
Node placementglobal.nodeSelector.hostname<component>.nodeSelector.hostnameSchedule all pods on the same node.
Image pull secretglobal.image.secret<component>.image.secretSingle docker-registry secret name for nvcr.io across all components.
Multus networkglobal.network.name<component>.network.nameNetworkAttachmentDefinition name; must match for ST 2110 traffic.
Schedulerglobal.schedulerName<component>.schedulerNamePod spec.schedulerName for all components.

Use global values when deploying the umbrella chart so that node selector, pull secret, network, and scheduler need to be specified only once instead of per component.


Cluster and Workload

ConfigurationHelm KeyComment
Deploy NIM workloadenabledDefault: chart-dependent; set false to skip NIM (for example, partial demo chart installation).
Name prefixappNameDeployment and pod naming.
ReplicasreplicasDefault: 1 for GPU media functions.
Node placementnodeSelector.hostnameMatches kubernetes.io/hostname (or cluster labels).
SchedulerschedulerNameDefault: topo-aware-scheduler for NUMA-aligned placement of the GPU, SR-IOV virtual function, and CPU pinning. Override on clusters that lack a topology-aware scheduler (for example, set to default-scheduler); the umbrella global.schedulerName overrides this setting for all components when non-empty. Refer to Scheduler and NUMA Topology.
Container imageimage.repository, image.tagPublished NIM image.
Image pull secretimage.secretdocker-registry secret name for nvcr.io.
Container securitycontainerSecurityContext.*runAsUser, runAsGroup, runAsNonRoot, allowPrivilegeEscalation, capabilities.
Multus network attachmentnetwork.nameDefault: media-a-tx-net; must match your NetworkAttachmentDefinition name.

NGC and Model Download

ConfigurationHelm KeyComment
Model download secretngc.secretNameSecret containing NGC_API_KEY.
Secret key namengc.secretKeyDefault: NGC_API_KEY.

Model Cache and Logs PVCs

ConfigurationHelm KeyComment
Model cache PVCnimModelCache.*Default path: /opt/nim/.cache.
NIM logs PVCnimLogs.*Default path: /workspace/nim-logs.

Subkeys: enabled, create, size, storageClassName, mountPath, readOnly, subPath. For configuration examples, refer to Advanced Usage.


NMOS

ConfigurationHelm KeyEnvironment VariableComment
NMOS discovery modenmos.enabled—true or false; false requires SMPTE ST 2110 static only (fixed IP addresses, ports, sessions). The actual pipeline front-end is controlled by studiovoice.nmosEnabled in Pipeline Configuration.
NMOS HTTP portnmos.httpPortNVDS_NMOS_DEFAULT_HTTP_PORTDefault: 9010.
NMOS seednmos.seedNVDS_NMOS_DEFAULT_SEEDNMOS node seed identifier.
Advertised hostnamenmos.hostnameSTUDIOVOICE_NMOS_HOSTNAMEmDNS / NMOS IS-04.
Descriptionnmos.descriptionSTUDIOVOICE_NMOS_DESCRIPTIONNMOS description string.
Labelnmos.labelSTUDIOVOICE_NMOS_LABELNMOS label string.

NMOS HTTP port 9010 and the Multus network attachment media-a-tx-net are often set in chart templates—confirm for your chart version.


Service Endpoints

The NIM service chart and umbrella sample chart create a Kubernetes Service for each component that exposes the NMOS Node API (IS-04/IS-05) on port 9010. Use these to reach the NMOS Connection Manager from outside the pod without kubectl port-forward.

ConfigurationHelm KeyComment
Expose endpointservice.enabledDefault: true. Set false to skip Service creation.
Service typeservice.typeDefault: NodePort. Supported: ClusterIP, NodePort, LoadBalancer. Use ClusterIP for in-cluster access only and LoadBalancer when the cluster has an external load balancer.
Service portservice.portDefault: 9010 (matches nmos.httpPort).
NodePortservice.nodePortUsed only when service.type: NodePort. Defaults: 32510 (sender), 32511 (receiver), 32512 (NIM service).

In the umbrella sample chart, the same keys apply per-component under sender.service.*, receiver.service.*, and nvidia-studio-voice-h4m-service.service.*. The Service name is <appName>-endpoint.

Note that the NIM HTTP API endpoints described in Observability (port 8000, /v1/health/*, /v1/metrics, and so on) are exposed inside the container and require kubectl port-forward for external access; they are not fronted by the chart-managed Service.


See Also