How-to | Use a Sandbox node to test workflows on a new Dataiku version

Important

The Sandbox node isn’t available in all Dataiku plans. Reach out to your Dataiku representative to learn more.

To minimize post-upgrade disruptions, Dataiku Cloud offers a Sandbox node for testing workflows on a new version of the software.

Activate the Sandbox node

To get started, first install the extension.

1. Navigate to the Extensions panel of the Launchpad.

2. Click + Add an Extension.

3. Select Sandbox node.

4. By default, the designers group has access to the Sandbox node. Update access rights according to your policy, as described in permission management on Dataiku Cloud.

Attention

If the IPsec extension is already active when you create the Sandbox node, IPsec isn’t enabled automatically on that node. To enable IPsec for the Sandbox node, edit the IPsec extension configuration and add the Sandbox node. This change triggers a restart of all nodes.

Identify dependencies of representative projects

The next step is to identify at least one representative Design node project to use for testing the new Dataiku version. What qualifies as a representative project may differ per organization, but it likely includes some combination of the following:

• At least one custom code environment

• Connections to external data sources

• Messaging channels for Reporting on scenario runs

After choosing a representative project (or projects), create the list of dependencies.

Important

If the project(s) have Shared objects, identify the source projects, and add them to the list of projects to import onto the Sandbox node.

Create dependencies on the Sandbox node

Before importing any representative projects onto the Sandbox node, first create the dependencies on the Sandbox node.

1. From the Overview panel of the Launchpad, as a user with admin rights, open the Sandbox node.

2. As you would on any Dataiku instance, create the connections identified in the previous step, using the same name when possible. This avoids needing to remap the connection during project import. Update the connection configuration if needed.

3. Similarly, create and build the code environments identified in the previous step by copying the Packages to install from the associated code environment in your Design node. As with connections, using the same name avoids code environment remapping during project import.

4. If needed, create the messaging channels identified in the previous step.

Attention

Be careful when replicating connection configurations. To avoid overwriting data from the main Design node with Sandbox node workflows, you need to update the path / prefix in each connection. For instance, for database connections, you can update the schema / table prefix to ${NODE}_${projectKey}_.

Tip

Managed connections, such as dataiku-managed-storage, are automatically created on the Sandbox node with a different per-node path prefix. This separates data when dataset names and project keys are identical on the Design node and the Sandbox node.

Import representative projects

With their dependencies prepared, you can now import the selected projects. Using a Designer-like profile, for each project identified previously:

1. In the Design node, export the project, and download the ZIP file.

2. In the Sandbox node, go to + New Project > Import project. Select the ZIP file downloaded in the previous step, and click Import.

3. Apply the required remapping based on differences in connection and code environment names. If you used the same names, the import UI skips the remapping step.

4. Trigger a workflow build (for example, through a scenario run), and check the results.

If your project produces the expected output, you’ve confirmed that upgrading Dataiku doesn’t affect its behavior.

Tip

Dataiku Cloud propagates containerized execution configurations to all nodes, so remapping isn’t necessary for those items.

Upgrade the Sandbox node

The Sandbox node can follow a different version lifecycle than your other nodes.

1. To upgrade the Sandbox node, go to the Upgrade and maintenance preferences section on the Launchpad Settings panel.

2. If a new Dataiku version is available, the UI proposes scheduling the upgrade for all nodes or for Sandbox nodes.

3. Select the latter, then choose to upgrade now or schedule the upgrade in the future.

4. After the upgrade finishes, the Launchpad Overview panel displays the Sandbox node on a different version than the rest of the space.

5. Trigger your workflow build, and check the results.

Tip

If some code environments fail to rebuild during the upgrade, Dataiku Cloud sends a notification email to space administrators.