Skip to navigation

Getting Started

View as MarkdownOpen in Claude

Holoscan for Media

Platform setup, cluster prerequisites, reference applications, deployment, verification, monitoring, application development, troubleshooting, and support for Holoscan for Media are covered in the NVIDIA Holoscan for Media user guide.


Prerequisites

Before proceeding, check the Support Matrix to verify that the required hardware and software are available.

RequirementDetailsReference
Kubernetes clusterKubernetes or Red Hat OpenShift.Support Matrix
OpenShift setupCreate the target namespace and grant SCC permissions to each service account your workloads use, often default.OpenShift Security Context Constraint
Namespace scopeNetworkAttachmentDefinition resources are namespace-scoped. Copy them into the deployment namespace when they exist in another namespace.Namespace Scope
NGC account and API keyRequired for CLI authentication, image and chart pulls, and creating image pull and model pull secrets.NGC Authentication
Rivermax licenseRequired for SMPTE ST 2110 streaming.Rivermax License
High-speed network attachmentA NetworkAttachmentDefinition for SMPTE ST 2110 provisioned by the cluster administrator.High-Speed Network Configuration
OpenShift Security Context ConstraintAn SCC that grants the NIM Linux capabilities and allowPrivilegeEscalation: true.Security Context
Topology-aware schedulerRequired on clusters with topologyManagerPolicy: single-numa-node.Scheduler and NUMA Topology
NMOS registryRequired only when using NMOS for dynamic stream connection management.-
Chrome Remote DesktopRequired to reach the NMOS Connection Manager UI in Holoscan for Media setups.Chrome Remote Desktop

Access NGC and Helm Charts

An NGC account and an NGC API key are required to authenticate the CLI, pull container images and Helm charts from NVIDIA registries, and create cluster pull secrets.

NGC Authentication

Generate an NGC API key at https://org.ngc.nvidia.com/setup/api-keys. Ensure that at least NGC Catalog is selected from the Services Included dropdown.

Export the key:

export NGC_API_KEY=<value>

Log in to the NGC container and Helm registry:

helm registry login nvcr.io --username '$oauthtoken' --password $NGC_API_KEY

License and Secrets Management

Rivermax License

Rivermax is required by Holoscan for Media applications to comply with SMPTE ST 2110 timing and traffic-flow requirements.

Create a Kubernetes secret from the license file:

kubectl create secret generic rivermax-license \
--from-file=rivermax.lic \
-n <your-namespace>

On Red Hat OpenShift, replace kubectl with oc.

Image Pull Secret

Create a docker-registry secret so the cluster can pull images from nvcr.io:

kubectl create secret docker-registry <image-pull-secret> \
--docker-server=nvcr.io \
'--docker-username=$oauthtoken' \
--docker-password=$NGC_API_KEY \
--docker-email=<your-email> \
-n <your-namespace>

Model Pull Secret

Model artifacts are downloaded by the NIM at runtime using an API key carried in a generic secret, separate from the image pull secret:

kubectl create secret generic <model-pull-secret> \
--from-literal=NGC_API_KEY=$NGC_API_KEY \
-n <your-namespace>

High-Speed Network Configuration

Set the network attachment name in the Helm values for each component in the pipeline:

network:
name: media-a-tx-net

Namespace Scope

NetworkAttachmentDefinition resources are namespace-scoped, and Multus looks them up in the same namespace as the pod. Every network attachment referenced by the chart must exist in the namespace where BNR is installed.

Check which network attachments already exist in your deployment namespace:

kubectl get net-attach-def -n <your-namespace>

On Red Hat OpenShift, replace kubectl with oc.

If a referenced network attachment is missing, pod admission fails with an error such as could not find network attachment definition '<namespace>/media-a-tx-net'.

Security Context

To operate correctly, the NIM container requires allowPrivilegeEscalation: true and the following Linux capabilities:

CapabilityPurpose
IPC_LOCKLock memory pages; required for hugepage allocation by Rivermax.
NET_RAWUse raw sockets; required by Rivermax for SMPTE ST 2110.
SYS_NICESet process and thread CPU affinity; required by Rivermax.
DAC_READ_SEARCHRead files across permission boundaries during startup.

These values are set in the Helm values or in the BNR operator custom resource under spec.parameters.securityContext. For details, refer to Configuration.

OpenShift Security Context Constraint

For Red Hat OpenShift, grant an SCC that allows the following capabilities:

  • IPC_LOCK
  • NET_RAW
  • SYS_NICE
  • DAC_READ_SEARCH
  • allowPrivilegeEscalation: true

For development clusters, administrators often grant the built-in privileged SCC to the service account used by the release:

oc adm policy add-scc-to-user privileged -z default -n <your-namespace>

Scheduler and NUMA Topology

The chart defaults to schedulerName: topo-aware-scheduler. This keeps the GPU, SR-IOV virtual function, and CPU allocation NUMA-aligned on clusters that enforce the single-numa-node topology manager policy.

To override schedulerName on clusters without a topology-aware scheduler, use the matching knob for your deployment path:

--set global.schedulerName=default-scheduler

Chrome Remote Desktop

When BNR is deployed with NMOS enabled, streams are connected and managed through the NMOS Connection Manager web UI. On Holoscan for Media clusters, access to that UI typically requires a browser that can reach the cluster network.