Skip to content

Installing

⚠ Warning

The instructions for installing SAME CLI binary tool will use the version generated from the archived https://github.com/azure-octo/same-cli private repo.

To use the updated same-mono-private version of the tool, you will need to run it from source code as described in How to set up your environment for running tests and building SAME.

Installing SAME

In order to use the SAME CLI to build and deploy your code, you first need to install the SAME binary. This is a no-dependency binary that runs on various operating systems, with pre-built binaries for recent versions of Windows, MacOS and Linux.

{{< tabs From-Source Linux MacOS Windows Binaries >}} {{% codetab %}}

It is currently recommended to clone the repo and run SAME CLI directly from the Python virtual environment.

For details, refer to How to set up your environment for running tests and building SAME.

{{% /codetab %}}

{{% codetab %}}

Execute the following command to install SAME on Linux.

curl -L0 https://get.sameproject.org/ | bash -

{{% /codetab %}}

{{% codetab %}}

Execute the following command to install SAME on MacOS.

curl -L0 https://get.sameproject.org/ | bash -

{{% /codetab %}}

{{% codetab %}}

# Not supported

{{% /codetab %}}

{{% codetab %}}

  1. Download the same CLI from the pre-built packages and signatures:

  2. https://github.com/SAME-Project/SAMPLE-CLI-TESTER/releases

  3. Ensure the user has permission to execute the binary and place it somewhere on your PATH so it can be invoked easily.

{{% /codetab %}} {{< /tabs >}}

Connecting to a Workflow engine

To run SAME, you will need a workflow engine to connect to. We support a variety of workflow engines, but recommend that, for now, you connect to one that is dedicated to SAME exclusively. Below are installation and configuration instructions for each.

{{< tabs Kubernetes-Kubeflow Azure-Machine-Learning Sagemaker Airflow >}}

{{% codetab %}}

  1. Verify that you have a Kubernetes cluster configured with a local kubectl context set as the default. To verify this, run kubectl config current-context and verify that it returns the name of your cluster.

  2. Install Kubeflow on Kubernetes. We recommend using a Terrachain from Combinator.ml. https://combinator.ml/stacks/kubeflow-mlflow//

Note: We support Kubernetes in Docker as well, though this can be quite resource intensive. You may wait up to 10 minutes for the service to become available after the first installation.

  1. Ensure your $KUBECONFIG is set. SAME can operate without this, but some tools may expect this to be set.
set KUBECONFIG="~/.kube/config"

{{% /codetab %}}

{{% codetab %}}

In order to use Azure Machine Learning as a target, you will need to:

  1. Authorize with Azure
  2. Set a series of environment variables that describe your workspace
  3. Provision the necessary compute ahead of time.

We will walk through all of these now.

Authorizing with Azure

First, you should download and install the Azure CLI - https://docs.microsoft.com/en-us/cli/azure/install-azure-cli

You will also need to create a service principal for accessing the AML workspace. You can follow the instructions for doing that here. https://docs.microsoft.com/en-us/azure/machine-learning/how-to-manage-workspace-cli?tabs=createnewresources%2Cvnetpleconfigurationsv1cli

Setting Your Environment Variables

After executing this, you will get a JSON blob output that you will need to set as environment variables. You will need to set all of the following:

export AML_SP_NAME="xxxx"
export AML_SP_APP_ID="<UID>"
export AML_SP_OBJECT_ID="<UID>"
export AML_SP_TENANT_ID="<UID>"
export AML_SP_PASSWORD_VALUE="**PASSWORD**"
export AML_SP_PASSWORD_ID="<UID>"

export WORKSPACE_SUBSCRIPTION_ID="<UID>"
export WORKSPACE_RESOURCE_GROUP="Resource_Group_Name"
export WORKSPACE_NAME="WORKSPACE_NAME"

Setting Up Compute cluster

Finally, you will need to setup your compute cluster(s) to target with SAME. The instructions to do that are here: https://docs.microsoft.com/en-us/azure/machine-learning/how-to-create-attach-compute-cluster?tabs=azure-cli

{{% /codetab %}}

{{% codetab %}}

-- Not Implemented --

{{% /codetab %}}

{{% codetab %}}

-- Not Implemented --

{{% /codetab %}}

{{< /tabs >}}

Next

You're done setting up SAME and are now ready to execute! The default execution uses Kubeflow (either locally or in the cloud). Use the -t flag to set another target.

Back to top