A Helm chart for deploying NVIDIA Device Plugin DaemonSet to allow scheduling of GPU workloads on nodes with NVIDIA GPUs.
| Name | Url | |
|---|---|---|
| la-cc | artem@lajko.dev | https://github.com/la-cc |
To install the chart with the release name nvidia-device-plugin:
$ helm repo add nvidia-device-plugin https://la-cc.github.io/nvidia-device-plugin-helm-chart
$ helm repo update
$ helm install nvidia-device-plugin nvidia-device-plugin/nvidia-device-plugin| Key | Type | Default | Description |
|---|---|---|---|
| env | list | [{"name":"FAIL_ON_INIT_ERROR","value":"false"}] |
Environment variables to set in the container. |
| env[0] | object | {"name":"FAIL_ON_INIT_ERROR","value":"false"} |
Name of the environment variable. |
| env[0].value | string | "false" |
Value of the environment variable. |
| fullnameOverride | string | "nvidia-device-plugin" |
Override the full name of the chart. |
| image | object | {"pullPolicy":"IfNotPresent","repository":"nvcr.io/nvidia/k8s-device-plugin","tag":"v0.16.1"} |
Configuration for the Docker image used by the pod. |
| image.pullPolicy | string | "IfNotPresent" |
The pull policy for the image. IfNotPresent means the image will only be pulled if it is not already present locally. |
| image.repository | string | "nvcr.io/nvidia/k8s-device-plugin" |
The Docker repository to pull the image from. |
| image.tag | string | "v0.16.1" |
The specific tag of the image to use. |
| nodeSelector | object | {} |
Constraints to ensure pods are scheduled on nodes with specific attributes. |
| priorityClassName | string | "system-node-critical" |
Priority class name for the pod, indicating its scheduling priority. |
| replicaCount | int | 1 |
Number of pod replicas to deploy. |
| securityContext | object | {"allowPrivilegeEscalation":false,"capabilities":{"drop":["ALL"]}} |
Security options for the pod. |
| securityContext.allowPrivilegeEscalation | bool | false |
Prevent privilege escalation within the container. |
| securityContext.capabilities | object | {"drop":["ALL"]} |
Capabilities to add or drop from the default set. |
| securityContext.capabilities.drop | list | ["ALL"] |
Drop all capabilities for the container. |
| tolerations | list | [{"effect":"NoSchedule","key":"ai","operator":"Equal","value":"true"}] |
Tolerations for taints on nodes, allowing pods to be scheduled on tainted nodes. |
| tolerations[0] | object | {"effect":"NoSchedule","key":"ai","operator":"Equal","value":"true"} |
The taint key that the toleration applies to. |
| tolerations[0].effect | string | "NoSchedule" |
The effect of the taint; pods with this toleration can be scheduled on nodes with the taint. |
| tolerations[0].operator | string | "Equal" |
The operator to use for the toleration. |
| tolerations[0].value | string | "true" |
The value of the taint to tolerate. |
| volumeMounts | list | [{"mountPath":"/var/lib/kubelet/device-plugins","name":"device-plugin"}] |
Volumes to mount into the container. |
| volumeMounts[0] | object | {"mountPath":"/var/lib/kubelet/device-plugins","name":"device-plugin"} |
Name of the volume. |
| volumeMounts[0].mountPath | string | "/var/lib/kubelet/device-plugins" |
Path inside the container where the volume should be mounted. |
| volumes | list | [{"hostPath":{"path":"/var/lib/kubelet/device-plugins"},"name":"device-plugin"}] |
Volumes to be used by the pod. |
| volumes[0] | object | {"hostPath":{"path":"/var/lib/kubelet/device-plugins"},"name":"device-plugin"} |
Name of the volume. |
| volumes[0].hostPath | object | {"path":"/var/lib/kubelet/device-plugins"} |
Use a directory from the host node's filesystem. |
| volumes[0].hostPath.path | string | "/var/lib/kubelet/device-plugins" |
Path on the host node to use as the volume. |
Autogenerated from chart metadata using helm-docs v1.14.2