Kubeflow Orchestrator

How to orchestrate pipelines with Kubeflow

This is an older version of the ZenML documentation. To read and view the latest version please visit this up-to-date URL.

The Kubeflow orchestrator is an orchestrator flavor provided with the ZenML kubeflow integration that uses Kubeflow Pipelines to run your pipelines.

This component is only meant to be used within the context of remote ZenML deployment scenario. Usage with a local ZenML deployment may lead to unexpected behavior!

When to use it

You should use the Kubeflow orchestrator if:

  • you're looking for a proven production-grade orchestrator.

  • you're looking for a UI in which you can track your pipeline runs.

  • you're already using Kubernetes or are not afraid of setting up and maintaining a Kubernetes cluster.

  • you're willing to deploy and maintain Kubeflow Pipelines on your cluster.

How to deploy it

The Kubeflow orchestrator supports two different modes: Local and remote. In case you want to run the orchestrator on a local Kubernetes cluster running on your machine, there is no additional infrastructure setup necessary.

If you want to run your pipelines on a remote cluster instead, you'll need to set up a Kubernetes cluster and deploy Kubeflow Pipelines:

  • Have an existing AWS EKS cluster set up.

  • Make sure you have the AWS CLI set up.

  • Download and install kubectl and configure it to talk to your EKS cluster using the following command:

    aws eks --region REGION update-kubeconfig --name CLUSTER_NAME
  • Install Kubeflow Pipelines onto your cluster.

If one or more of the deployments are not in the Running state, try increasing the number of nodes in your cluster.

If you're installing Kubeflow Pipelines manually, make sure the Kubernetes service is called exactly ml-pipeline. This is a requirement for ZenML to connect to your Kubeflow Pipelines deployment.

Infrastructure Deployment

A Kubeflow orchestrator can be deployed directly from the ZenML CLI:

zenml orchestrator deploy kubeflow_orchestrator --flavor=kubeflow ...

You can pass other configuration specific to the stack components as key-value arguments. If you don't provide a name, a random one is generated for you. For more information about how to work use the CLI for this, please refer to the dedicated documentation section.

How to use it

To use the Kubeflow orchestrator, we need:

  • The ZenML kubeflow integration installed. If you haven't done so, run

    zenml integration install kubeflow
  • Docker installed and running.

  • kubectl installed.

When using the Kubeflow orchestrator locally, you'll additionally need:

  • K3D installed to spin up a local Kubernetes cluster.

  • Terraform installed to set up the Kubernetes cluster with various deployments.

To run the pipeline on a local Kubeflow Pipelines deployment, you can use the ZenML Stack recipes to spin up a local Kubernetes cluster and install Kubeflow Pipelines on it. The stack recipe is called k3d-modular and is available in the ZenML stack recipe repository. The recipe is modular, meaning that you can configured it to use different orchestrators, Model Deployers, and other tools.

To deploy the stack, run the following commands:

# Pull the `k3d-modular` recipe to your local system
zenml stack recipe pull k3d-modular
# Deploy the stack using the ZenML CLI:
zenml stack recipe deploy k3d-modular -i kubeflow -i minio --no-server
# run the following command to import the resources as a ZenML stack, manually
# set the imported stack as the active stack
zenml stack set <STACK_NAME>
# Get the Kubeflow Pipelines UI endpoint
kubectl get ingress -n kubeflow  -o jsonpath='{.items[0].spec.rules[0].host}'

You can read more about the recipes in Stack Recipes. or check the recipe code in the ZenML Stack Recipe Repository.

The local Kubeflow Pipelines deployment requires more than 4 GB of RAM, and 30 GB of disk space, so if you are using Docker Desktop make sure to update the resource limits in the preferences.

ZenML will build a Docker image called <CONTAINER_REGISTRY_URI>/zenml:<PIPELINE_NAME> which includes your code and use it to run your pipeline steps in Kubeflow. Check out this page if you want to learn more about how ZenML builds these images and how you can customize them.

You can now run any ZenML pipeline using the Kubeflow orchestrator:

python file_that_runs_a_zenml_pipeline.py

Kubeflow UI

Kubeflow comes with its own UI that you can use to find further details about your pipeline runs, such as the logs of your steps. For any runs executed on Kubeflow, you can get the URL to the Kubeflow UI in Python using the following code snippet:

from zenml.post_execution import get_run

pipeline_run = get_run("<PIPELINE_RUN_NAME>")
orchestrator_url = deployer_step.metadata["orchestrator_url"].value

Additional configuration

For additional configuration of the Kubeflow orchestrator, you can pass KubeflowOrchestratorSettings which allows you to configure (among others) the following attributes:

  • client_args: Arguments to pass when initializing the KFP client.

  • user_namespace: The user namespace to use when creating experiments and runs.

  • pod_settings: Node selectors, affinity and tolerations to apply to the Kubernetes Pods running your pipeline. These can be either specified using the Kubernetes model objects or as dictionaries.

from zenml.integrations.kubeflow.flavors.kubeflow_orchestrator_flavor import KubeflowOrchestratorSettings
from kubernetes.client.models import V1Toleration

kubeflow_settings = KubeflowOrchestratorSettings(
        "affinity": {
            "nodeAffinity": {
                "requiredDuringSchedulingIgnoredDuringExecution": {
                    "nodeSelectorTerms": [
                            "matchExpressions": [
                                    "key": "node.kubernetes.io/name",
                                    "operator": "In",
                                    "values": ["my_powerful_node_group"],
        "tolerations": [

        "orchestrator.kubeflow": kubeflow_settings

Check out the API docs for a full list of available attributes and this docs page for more information on how to specify settings.

Enabling CUDA for GPU-backed hardware

Note that if you wish to use this orchestrator to run steps on a GPU, you will need to follow the instructions on this page to ensure that it works. It requires adding some extra settings customization and is essential to enable CUDA for the GPU to give its full acceleration.

Important Note for Multi-Tenancy Deployments

Kubeflow has a notion of multi-tenancy built into its deployment. Kubeflow's multi-user isolation simplifies user operations because each user only views and edited\s the Kubeflow components and model artifacts defined in their configuration.

Using the ZenML Kubeflow orchestrator on a multi-tenant deployment without any settings will result in the following error:

HTTP response body: {"error":"Invalid input error: Invalid resource references for experiment. ListExperiment requires filtering by namespace.","code":3,"message":"Invalid input error: Invalid resource references for experiment. ListExperiment requires filtering by 
namespace.","details":[{"@type":"type.googleapis.com/api.Error","error_message":"Invalid resource references for experiment. ListExperiment requires filtering by namespace.","error_details":"Invalid input error: Invalid resource references for experiment. ListExperiment requires filtering by namespace."}]}

In order to get it to work, we need to leverage the KubeflowOrchestratorSettings referenced above. By setting the namespace option, and by passing in the right authentication credentials to the Kubeflow Pipelines Client, we can make it work.

First, when registering your kubeflow orchestrator, please make sure to include the kubeflow_hostname parameter. The kubeflow_hostname must end with the /pipeline post-fix.

zenml orchestrator register <NAME> \
    --flavor=kubeflow \
    --kubernetes_context=<KUBERNETES_CONTEXT> \  
    --kubeflow_hostname=<KUBEFLOW_HOSTNAME> # e.g. https://mykubeflow.example.com/pipeline

Then, ensure that you use the pass the right settings before triggering a pipeline run. The following snippet will prove useful:

import requests

from zenml.client import Client
from zenml.integrations.kubeflow.flavors.kubeflow_orchestrator_flavor import (

NAMESPACE = "namespace_name"  # This is the user namespace for the profile you want to use
USERNAME = "admin"  # This is the username for the profile you want to use
PASSWORD = "abc123"  # This is the password for the profile you want to use

# Use client_username and client_password and ZenML will automatically fetch a session cookie
kubeflow_settings = KubeflowOrchestratorSettings(

# You can also pass the cookie in `client_args` directly
# kubeflow_settings = KubeflowOrchestratorSettings(
#     client_args={"cookies": session_cookie}, user_namespace=NAMESPACE
# )

        "orchestrator.kubeflow": kubeflow_settings

if "__name__" == "__main__":
  # Run the pipeline

Note that the above is also currently not tested on all Kubeflow versions, so there might be further bugs with older Kubeflow versions. In this case, please reach out to us on Slack.

Using secrets in settings

The above example encoded the username and password in plain-text as settings. You can also set them as secrets.

zenml secret create kubeflow_secret \
    --username=admin \

And then you can use them in code:

# Use client_username and client_password and ZenML will automatically fetch a session cookie
kubeflow_settings = KubeflowOrchestratorSettings(
    client_username="{{kubeflow_secret.username}}",  # secret reference
    client_password="{{kubeflow_secret.password}}",  # secret reference

See full documentation of using ZenML secrets here.

A concrete example of using the Kubeflow orchestrator can be found here.

For more information and a full list of configurable attributes of the Kubeflow orchestrator, check out the API Docs.

Last updated