Non-Default Namespaces

Often, production deploys of HPE Machine Learning Data Management involve deploying HPE Machine Learning Data Management to a non-default namespace. This helps administrators of the cluster more easily manage HPE Machine Learning Data Management components alongside other things that might be running inside of Kubernetes (DataDog, TensorFlow Serving, etc.).

• To deploy HPE Machine Learning Data Management to a non-default namespace, you need to add the -n or --namespace flag when deploying. If the namespace does not already exist, you can have Helm create it with --create-namespace.

  helm install <args> --namespace pachyderm --create-namespace

  Copy

• To talk to your HPE Machine Learning Data Management cluster:

  • You can either modify an existing pachctl context

    pachctl config update context --namespace pachyderm

    Copy

  • or import one from Kubernetes: