Guided setup 1: Deploy in a new VNet with Elastic Compute¶

Description ¶

This guided setup allows you to setup a full Dataiku Cloud Stacks for Azure setup, including the ability to run workloads on Elastic Compute clusters powered by Kubernetes (using Azure AKS).

At the end of this setup, you’ll have:

A fully-managed DSS design node, with either a public IP or a private one
The ability to one-click create elastic compute clusters
The elastic compute clusters running with public IPs (and no NAT gateway overhead)

You need to have, as an administrator, the Owner role on the resource group. Ownership is required to give permissions to the managed identities used by the software at runtime. Said identities do not require to have the role Owner themselves. Owner privileges are only used by the administrator for initial setup and is not required by the software during usage.

For the following steps, we assume an ACR (Azure Container Registry) is created and available in your resource group.

Steps ¶

Create a managed identity for your Fleet Manager instance ¶

In your Microsoft Azure portal, click on “Create a resource”
Search for “User Assigned Managed Identity”
Click on create
Select the correct subscription and resource group
Select the region
Enter a name for your managed identity, this will be referred as fm-id-name
Click on “Review + Create”, then on “Create”
When your deployment is ready, click on “Go to resource”
Click on “Azure role assignments”, then “Add role assignment”
For the “Scope”, select “Resource group”
Select the correct subscription and resource group
For the “Role” select “Contributor”
Click on “Save”
Go to the “Properties” tab and copy the “Resource ID”, this will be refered as fm-id

Create a managed identity for your DSS instances ¶

Reproduce the exact same step as above (i.e. the steps for the Fleet Manager managed identity) for the DSS managed identity. We will refer to its name by dss-id-name and to its “Resource ID” by dss-id

As a consequence of role assignments inheritance, this identity is also Contributor on the ACR.

Deploy Fleet Manager ¶

Click the following link to deploy the Fleet Manager and all needed resources from the Azure portal:

In the portal fill the required information for the deployment:

Choose the target subscription
Choose an existing resource group or create a new one
Select the region where the resource should be deployed
Select the size of the Virtual machine
In “Instance identity”, enter the fm-id-name
In “Username”, choose a username for logging in to Fleet Manager
In “Password”, enter a strong password for logging in to Fleet Manager
In “SSH Username”, choose a username for logging in to the underlying Fleet Manager virtual machine (it is not normally required)
In “SSH public key source”, enter a RSA public key for SSH logging in to the underlying Fleet Manager virtual machine (it is not normally required)
In “Virtual Network CIDR”, enter a /16 CIDR, for example 10.0.0.0/16
In “Subnet CIDR”, enter a CIDR included in the chosen Virtual Network CIDR, for example 10.0.1.0/24
In “Secondary subnet CIDR”, enter a CIDR included in the chosen Virtual Network CIDR that does not overlap with the previously chosen Subnet CIDR, for example 10.0.2.0/24
In “Associate Public Ip Address”, select whether you want a public IP to connect to Fleet Manager
Click on “Review + create”
Verify the creation information, then click on “Create”
Wait for your deployment to appear as completed
Click on “Go to resource group”
Click on the Virtual Machine
Copy the “Public IP address”

This is the address at which your Cloud Stacks Fleet manager is deployed. Open a new tab to this address and wait for the login screen to appear.

Start your first DSS ¶

Log into Fleet Manager with the login and the password you previously entered
In “Cloud Setup”, click on “Edit”, set “License mode” to “Manually entered”, click on “Enter license” and enter your Dataiku license. Click on “Done” then on “Save”
Refresh the page in your browser
In “Fleet Blueprints”, click on “Elastic Design”
Give a name to your new fleet, this will be referred to as fleet-name
In “Authorized SSH Key”, enter your RSA public key
In “Instance managed identity”, enter the dss-id
Click on “Deploy”
Go to “Instances > All”, click on the design node
Click “Provision”
Wait for your DSS instance to be ready
Click on “Retrieve password” and write-down the password
Click on “Go to DSS”
Login with “admin” as the login, and the password you just retrieved

You can now start using DSS

(Optional) Start your first Elastic compute cluster ¶

Deploy your Elastic Compute cluster ¶

In DSS, go to Administration > Clusters
Click on “Create AKS cluster”, give it a name
In “Credentials”, check it is set to “Manually defined” -> “Default credentials, from environment”
In section “Identity assumed by cluster components”, check it is set to “Manually defined” -> “Managed identities”. The option “Inherit DSS identity” should be checked.
In “Cluster Nodes”, Click on “+ Add a preset”
Update “Machine type” and “Disk size” as you see fit
Tick the “Enable nodes autoscaling” box
Untick the “Availability zones” option in “Networking” section of the node pool.
In “Service CIDR” add a CIDR for your cluster, it should not overlap 10.0.0.0/16. For example: 10.1.0.0/16
In “DNS IP”, set an IP for the DNS, in your specified CIDR range. For example: 10.1.0.10
Click on “Start”
Wait for your cluster to be available
In Settings, go to “Containerized execution”, and in “Default cluster”, select the cluster you just created.
Click on “+ Add another config”
In “Configuration name” add a name
In “Image registry url” enter acr-name.azurecr.io
In “Image pre-push hook”, select “Enable push to ACR”
Click on “Save” then “Push base images”. When finished, click on “(Re)Install Jupyter kernels”.
In a project, you can now use containerized execution for any activity, using the containerized config you created

Troubleshooting ¶

“Tenant ID, application ID, principal ID, and scope are not allowed to be updated (code: RoleAssignmentUpdateNotPermitted)”¶

This error is known to happen whenever a new stack is deployed in a resource group in which an existing stack have been deleted. The most likely reason is some role assignments were left over and must be deleted before you can deploy again.

In your Microsoft Azure portal, display your resource group, then click on “Access Control (IAM)”
Select the “Role assignments” tab in the right blade.
Find all the entries with name “Identity not found” and scope “This resource” and select them.
Click on the “Remove” button at the top of the blade.
You can then redeploy the ARM template. No need to delete the resources which creation succeeded, just redeploy it like you did before the failure.

“The requested identity has not been assigned to this resource”¶

This error means you instructed DSS to authenticate with a specific User Assigned Managed Identity but this specific identity has not been assigned to the machine. Check your instance template for correct assignment of the desired identity, and reprovision if the error persists.