Cloudera Docs

Setting up the GPU node

In Kubernetes, you can taint nodes to affect how the node is scheduled. You can ensure that nodes that have a GPU are reserved exclusively for Cloudera AI workloads that require a GPU.

To reserve a GPU node, assign a taint to the node.

Assigning taint to the node on OpenShift

On OpenShift, specify the node taint nvidia.com/gpu: true:NoSchedule for any nodes that host GPUs and are required to be used only for GPU workloads.

Assigning taint to the node on Cloudera Embedded Container Service

On Cloudera Embedded Container Servicee, set the node taint nvidia.com/gpu: true:NoSchedule in one of the following three ways:

  1. During Cloudera Embedded Container Service installation

    After adding the GPU host(s) to Cloudera Manager but prior to creation of the Cloudera Embedded Container Service cluster:
    1. Visit the Host Configuration page.
    2. Select the Dedicated GPU Node for Data Services checkbox and Save the configuration.
    3. Repeat for all hosts on which the taint is desired.
    4. Proceed with installation via the Add Cluster wizard.

  2. During Cloudera Embedded Container Servicee upgrade

    After upgrading Cloudera Manager (if applicable):
    1. Set the host configuration as described in the first step above on one or more hosts in the Cloudera Embedded Container Service cluster.
    2. Proceed with upgrade via the Upgrade Cluster wizard.
  3. Independently of Cloudera Embedded Container Servicee install or upgrade
    1. Set the host configuration as described above on one or more hosts in the Cloudera Embedded Container Service cluster.
    2. Redeploy the client configuration on the Cloudera Embedded Container Service cluster.
    3. Complete a Rolling restart of the Cloudera Embedded Container Service cluster. Restart the Cloudera Embedded Container Service by selecting Cloudera Manager > Clusters > [***CLUSTER NAME***] > Actions.