Artifact Manifest

View as Markdown

This section provides a comprehensive list of all components required for NVIDIA Cloud Functions (NVCF) Self-Hosted deployment for basic inference. Additional components are needed for Low Latency Streaming (Simulation).

Artifacts Overview

The following inventories list the artifacts for an inference-only self-hosted NVCF deployment. Artifacts are grouped by deployment plane and type.

Artifact version compatibility

Newer artifact versions might be available. NVCF self-managed stack and compute-plane stack releases are QA-qualified as umbrella releases with the specific versions shown on this page. Use these versions together. NVIDIA cannot guarantee compatibility when you substitute other artifact versions.

Prepare Helm charts for Helmfile

The self-managed Helmfile bundles currently expect NVCF charts in an OCI registry. Chart distributions that start with https:// are Helm repository charts. Before deploying with Helmfile, copy each required repository chart version into the OCI registry configured by global.helm.sources in the Helmfile bundles.

The following example copies one public NVCF chart into an OCI registry:

$export CHART_NAME="helm-nvcf-api"
$export CHART_VERSION="1.23.9"
$export TARGET_REGISTRY="<registry-host>"
$export TARGET_REPOSITORY="<repository>"
$
$helm repo add nvcf https://helm.ngc.nvidia.com/nvidia/nvcf
$helm repo update
$helm pull "nvcf/${CHART_NAME}" --version "${CHART_VERSION}"
$
$helm registry login "${TARGET_REGISTRY}"
$helm push "${CHART_NAME}-${CHART_VERSION}.tgz" \
> "oci://${TARGET_REGISTRY}/${TARGET_REPOSITORY}"

Repeat this process for every required chart with an https:// distribution. Copy required charts with an nvcr.io distribution into the same target repository so Helmfile can resolve all NVCF charts from one source. Configure the stack environment with that OCI location:

1global:
2 helm:
3 sources:
4 registry: "<registry-host>"
5 repository: "<repository>"

See Image Mirroring for additional registry examples.

Use upstream container images

You can configure a chart to pull a supporting image directly from its upstream registry. For example, replace the nats.reloader.image block in deploy/stacks/self-managed/global.yaml.gotmpl to pull the NATS configuration reloader from Docker Hub:

1nats:
2 reloader:
3 image:
4 registry: docker.io
5 repository: natsio/nats-server-config-reloader
6 tag: "0.23.0"

Use the version listed in the artifact table. Verify that your cluster can reach the upstream registry. If the registry requires authentication, add its pull secret to global.imagePullSecrets.

To pull the API account-bootstrap Kubernetes utilities from their upstream image, replace the api.accountBootstrap.image block in global.yaml.gotmpl:

1api:
2 accountBootstrap:
3 image:
4 registry: docker.io
5 repository: alpine/k8s
6 tag: "1.36.1"

The current Cassandra initialization hook uses the nvcf-cassandra-migrations image, and the current NATS chart renders NKeys as Secrets without an nkey job. Their legacy cassandra.initialization.image and nats.nkeyJob.image values do not control rendered workloads. Using alpine-k8s for those operations requires chart support rather than a configuration-only override.

Some supporting components such as the GPU Operator, OpenBao, NATS, Cassandra, etc. can alternatively be pulled directly from public NGC Catalog or other public opensource repositories if desired.

The following tables list the complete artifact inventory.

Control plane Helm charts

ArtifactVersionRequiredDescriptionDistributionSource code
helm-admin-token-issuer-proxy1.4.3OptionalDeploys the admin token issuer proxy used by the reference architecture.https://helm.ngc.nvidia.com/nvidia/nvcf/helm-admin-token-issuer-proxy:1.4.3GitHub
helm-nvcf-api1.23.9RequiredDeploys the NVCF API service.https://helm.ngc.nvidia.com/nvidia/nvcf/helm-nvcf-api:1.23.9
helm-nvcf-api-keys1.6.0RequiredDeploys the API key management service.https://helm.ngc.nvidia.com/nvidia/nvcf/helm-nvcf-api-keys:1.6.0GitHub
helm-nvcf-cassandra0.18.0RequiredDeploys Cassandra and its initialization jobs.https://helm.ngc.nvidia.com/nvidia/nvcf/helm-nvcf-cassandra:0.18.0GitHub
helm-nvcf-cert-manager0.1.0RequiredDeploys the NVCF cert-manager configuration.https://helm.ngc.nvidia.com/nvidia/nvcf/helm-nvcf-cert-manager:0.1.0GitHub / Upstream
helm-nvcf-ess-api1.6.1RequiredDeploys the Encrypted Secrets Service API.https://helm.ngc.nvidia.com/nvidia/nvcf/helm-nvcf-ess-api:1.6.1
helm-nvcf-grpc-proxy1.6.7RequiredDeploys the gRPC proxy service.https://helm.ngc.nvidia.com/nvidia/nvcf/helm-nvcf-grpc-proxy:1.6.7GitHub
helm-nvcf-invocation-service1.5.6RequiredDeploys the HTTP invocation service.https://helm.ngc.nvidia.com/nvidia/nvcf/helm-nvcf-invocation-service:1.5.6GitHub
helm-nvcf-llm-api-gateway1.3.5OptionalDeploys the OpenAI-compatible LLM API gateway.nvcr.io/0833294136851237/nvcf-ncp-staging/helm-nvcf-llm-api-gateway:1.3.5GitHub
helm-nvcf-llm-request-router1.6.6OptionalDeploys the LLM request router.nvcr.io/0833294136851237/nvcf-ncp-staging/helm-nvcf-llm-request-router:1.6.6GitHub
helm-nvcf-nats0.7.1RequiredDeploys NATS messaging for the control plane.https://helm.ngc.nvidia.com/nvidia/nvcf/helm-nvcf-nats:0.7.1GitHub / Upstream
helm-nvcf-nats-auth-callout-service1.1.3RequiredDeploys the NATS authorization callout service.https://helm.ngc.nvidia.com/nvidia/nvcf/helm-nvcf-nats-auth-callout-service:1.1.3GitHub
helm-nvcf-notary-service1.4.2RequiredDeploys the notary service for signing and validation.https://helm.ngc.nvidia.com/nvidia/nvcf/helm-nvcf-notary-service:1.4.2
helm-nvcf-nvct-api1.4.3RequiredDeploys the NVCF tenant API service.https://helm.ngc.nvidia.com/nvidia/nvcf/helm-nvcf-nvct-api:1.4.3
helm-nvcf-openbao-server0.30.25RequiredDeploys OpenBao secret management.https://helm.ngc.nvidia.com/nvidia/nvcf/helm-nvcf-openbao-server:0.30.25GitHub / Upstream
helm-nvcf-rate-limiter1.0.3OptionalDeploys request rate limiting for supported invocation paths.https://helm.ngc.nvidia.com/nvidia/nvcf/helm-nvcf-rate-limiter:1.0.3GitHub
helm-nvcf-sis1.18.8RequiredDeploys the Spot Instance Service.https://helm.ngc.nvidia.com/nvidia/nvcf/helm-nvcf-sis:1.18.8
helm-nvcf-state-metrics1.0.1OptionalDeploys NVCF state metrics for observability.https://helm.ngc.nvidia.com/nvidia/nvcf/helm-nvcf-state-metrics:1.0.1
helm-nvcf-vanity-gateway0.1.0-nvcf-10204.1OptionalDeploys the optional vanity hostname gateway.https://helm.ngc.nvidia.com/nvidia/nvcf/helm-nvcf-vanity-gateway:0.1.0-nvcf-10204.1
helm-reval1.3.8RequiredDeploys the function revalidation service.https://helm.ngc.nvidia.com/nvidia/nvcf/helm-reval:1.3.8GitHub
nvcf-example-dashboards1.6.0OptionalDeploys example Grafana dashboards for NVCF telemetry.https://helm.ngc.nvidia.com/nvidia/nvcf/nvcf-example-dashboards:1.6.0
nvcf-gateway-routes1.14.0OptionalDeploys Gateway API routes for the reference architecture.https://helm.ngc.nvidia.com/nvidia/nvcf/nvcf-gateway-routes:1.14.0GitHub
nvcf-observability-reference-stack1.10.0OptionalDeploys a reference observability backend for evaluation.https://helm.ngc.nvidia.com/nvidia/nvcf/nvcf-observability-reference-stack:1.10.0

Control plane services and images

ArtifactVersionRequiredDescriptionDistributionSource code
admin-token-issuer-proxy1.0.2OptionalProxies admin token requests for the reference architecture.nvcr.io/nvidia/nvcf/admin-token-issuer-proxy:1.0.2
alpine-k8s1.36.1RequiredProvides Kubernetes command-line utilities for deployment jobs.docker.io/alpine/k8s:1.36.1GitHub
cassandra5.0.8-nv-2.0.1RequiredStores NVCF account, function, cluster, and service state.nvcr.io/nvidia/nvcf/cassandra:5.0.8-nv-2.0.1Upstream
cert-manager-cainjectorv1.20.2RequiredInjects certificate authority data into Kubernetes resources.nvcr.io/nvidia/nvcf/cert-manager-cainjector:v1.20.2Upstream
cert-manager-controllerv1.20.2RequiredReconciles certificates and issuers for the control plane.nvcr.io/nvidia/nvcf/cert-manager-controller:v1.20.2Upstream
cert-manager-startupapicheckv1.20.2RequiredVerifies that the cert-manager API is ready.nvcr.io/nvidia/nvcf/cert-manager-startupapicheck:v1.20.2Upstream
cert-manager-webhookv1.20.2RequiredValidates and converts cert-manager API resources.nvcr.io/nvidia/nvcf/cert-manager-webhook:v1.20.2Upstream
ess-apiv0.57.26RequiredProvides encrypted application secrets to NVCF workloads.nvcr.io/nvidia/nvcf/ess-api:v0.57.26
llm-api-gateway0.8.3OptionalExposes OpenAI-compatible APIs for LLM functions.nvcr.io/0833294136851237/selfhosted-ga/llm-api-gateway:0.8.3-eaGitHub
llm-request-router0.3.0OptionalRoutes LLM requests to eligible worker instances.nvcr.io/0833294136851237/selfhosted-ga/stargate:0.3.0-eaGitHub
nats-box0.19.7-nonrootRequiredProvides NATS administration and diagnostic utilities.nvcr.io/nvidia/nvcf/nats-box:0.19.7-nonrootUpstream
nats-server2.11.17-alpine3.22RequiredProvides messaging for function deployment and invocation.nvcr.io/nvidia/nvcf/nats-server:2.11.17-alpine3.22Upstream
nats-server-config-reloader0.23.0RequiredReloads NATS server configuration when mounted settings change.docker.io/natsio/nats-server-config-reloader:0.23.0Upstream
notary-service1.8.1RequiredSigns and validates functions and cluster nodes.nvcr.io/nvidia/nvcf/notary-service:1.8.1
nvcf-api-keys-service1.5.0RequiredCreates and manages NVCF API keys.nvcr.io/nvidia/nvcf/nvcf-api-keys-service:1.5.0
nvcf-cassandra-migrations0.10.3RequiredApplies the Cassandra schemas required by NVCF services.nvcr.io/0833294136851237/nvcf-ncp-staging/nvcf-cassandra-migrations:0.10.3GitHub
nvcf-grpc-proxy1.29.1RequiredProxies bidirectional gRPC traffic between the control and compute planes.nvcr.io/nvidia/nvcf/nvcf-grpc-proxy:1.29.1GitHub
nvcf-invocation-service0.8.8RequiredRoutes stateless HTTP function invocation requests.nvcr.io/nvidia/nvcf/nvcf-invocation-service:0.8.8GitHub
nvcf-nats-auth-callout-service0.5.10RequiredAuthorizes NATS clients for NVCF services and workloads.nvcr.io/nvidia/nvcf/nvcf-nats-auth-callout-service:0.5.10GitHub
nvcf-openbao2.5.5-nv-1.3.1RequiredStores and manages control-plane secrets.nvcr.io/0833294136851237/nvcf-ncp-staging/nvcf-openbao:2.5.5-nv-1.3.1Upstream
nvcf-openbao-migrations0.16.2RequiredApplies the OpenBao configuration required by NVCF.nvcr.io/nvidia/nvcf/nvcf-openbao-migrations:0.16.2GitHub
nvcf-service-oss1.9.0-hotfix.1RequiredProvides the primary NVCF control-plane API.nvcr.io/nvidia/nvcf/nvcf-service-oss:1.9.0-hotfix.1
nvct-service-oss1.5.9-hotfix.1RequiredProvides tenant-scoped NVCF control-plane operations.nvcr.io/nvidia/nvcf/nvct-service-oss:1.5.9-hotfix.1
oss-vault-k8s1.7.4RequiredIntegrates Kubernetes workloads with OpenBao secrets.nvcr.io/nvidia/nvcf/oss-vault-k8s:1.7.4
reval-server0.2.2RequiredRevalidates function state in the background.nvcr.io/nvidia/nvcf/reval-server:0.2.2GitHub
spot1.563.1-hotfix.1RequiredManages deployments, clusters, and function instances.nvcr.io/nvidia/nvcf/spot:1.563.1-hotfix.1

Compute plane Helm charts

ArtifactVersionRequiredDescriptionDistributionSource code
csi-driver-smbsupportedOptionalProvides SMB persistent volumes for supported deployments.https://raw.githubusercontent.com/kubernetes-csi/csi-driver-smb/master/chartsUpstream
ebs-csi-driversupportedOptionalProvides Amazon EBS persistent volumes for EKS clusters.https://kubernetes-sigs.github.io/aws-ebs-csi-driverUpstream
gpu-operatorsupportedRequiredManages NVIDIA GPU software on Kubernetes nodes.https://helm.ngc.nvidia.com/nvidiaUpstream
helm-nvca-operator1.12.7RequiredDeploys the NVCA operator and compute-plane integration.https://helm.ngc.nvidia.com/nvidia/nvcf/helm-nvca-operator:1.12.7GitHub
nvcf-container-cache0.25.22OptionalDeploys container image caching on GPU cluster nodes.https://helm.ngc.nvidia.com/nvidia/nvcf/nvcf-container-cache:0.25.22GitHub

Compute plane services and images

ArtifactVersionRequiredDescriptionDistributionSource code
ess-agent1.3.1RequiredInjects encrypted application secrets into function workloads.nvcr.io/nvidia/nvcf/ess-agent:1.3.1GitHub
gpu-operator-validatorsupportedRequiredValidates GPU Operator components on GPU nodes.https://catalog.ngc.nvidia.com/orgs/nvidia/teams/cloud-native/containers/gpu-operator-validatorUpstream
k8s-device-pluginsupportedRequiredAdvertises NVIDIA GPU resources to Kubernetes.https://catalog.ngc.nvidia.com/orgs/nvidia/teams/k8s/containers/device-pluginUpstream
nvca3.0.3RequiredRegisters GPU clusters and orchestrates deployments in-cluster.nvcr.io/nvidia/nvcf/nvca:3.0.3GitHub
nvca-operator3.0.3RequiredReconciles NVCA resources and compute-plane configuration.nvcr.io/nvidia/nvcf/nvca-operator:3.0.3
nvcf-container-cachev1.1.36OptionalCaches container image layers on GPU cluster nodes.nvcr.io/nvidia/nvcf/nvcf-container-cache:v1.1.36GitHub
nvcf-image-credential-helper0.10.2RequiredResolves container image credentials for function workloads.nvcr.io/nvidia/nvcf/nvcf-image-credential-helper:0.10.2GitHub
nvcf-proxy-tls-certsv1.2.10OptionalConfigures TLS trust for the optional container cache proxy.nvcr.io/nvidia/nvcf/nvcf-proxy-tls-certs:v1.2.10
nvcf-worker-init-oss1.0.9RequiredPrepares function resources before the user container starts.nvcr.io/0833294136851237/nvcf-ncp-staging/nvcf-worker-init-oss:1.0.9GitHub
nvcf-worker-llm-credentials-oss1.0.4OptionalMaintains a current NVCF worker token for LLM function workloads.nvcr.io/0833294136851237/nvcf-ncp-staging/nvcf-worker-llm-credentials-oss:1.0.4GitHub
nvcf-worker-utils-oss1.0.4RequiredProxies NATS traffic between function containers and the control plane.nvcr.io/0833294136851237/nvcf-ncp-staging/nvcf-worker-utils-oss:1.0.4GitHub
pylon0.3.2OptionalConnects LLM worker pods to the LLM request router.nvcr.io/nvidia/nvcf/pylon:0.3.2GitHub

Tools and deployment resources

ArtifactVersionDescriptionDistributionSource code
nvcf-cli1.10.3Manages functions, deployments, and clusters from the command line.nvcr.io/nvidia/nvcf/nvcf-cli:1.10.3GitHub
nvcf-compute-plane-stack1.0.6Provides the Helmfile bundle for compute-plane deployment.nvcr.io/nvidia/nvcf/nvcf-compute-plane-stack:1.0.6
nvcf-self-managed-stack0.6.1-rc.6Provides the Helmfile bundle for control-plane deployment.nvcr.io/nvidia/nvcf/nvcf-self-managed-stack:0.6.1-rc.6GitHub