Azure Kubernetes (AKS) Cluster Provisioning fails if the subnet(s) in use has an associated user-assigned NAT Gateway.
Incident Report for Cloudera
Resolved
Cloudera will publish a TSB to regarding this issue.
Posted May 25, 2022 - 13:16 UTC
Identified
Azure Kubernetes (AKS) Cluster Provisioning for CDP Data Services such as CML(Cloudera Machine Learning, CDE(Cloudera Data Engineering), and CDF(Cloudera Data Flow) will fail if the subnet(s) in use has an associated NAT Gateway.

Microsoft made a change starting from May, wherein creating an AKS cluster using the above Networking setup will fail unless the parameter: outboundType is set to userAssignedNATGateway.

The error is:
status: CREATE_FAILED message: '[error creating cluster] info: containerservice.ManagedClustersClient#CreateOrUpdate: Failure sending request: StatusCode=0 -- Original Error: Code="SubnetsAssociatedWithNATgatewayWhenOutboundTypeIsStandardLoadBalancer" Message="Subnets ''/subscriptions/017b8edd-1cb1-4c3e-aa2a-2f62880b1d3d/resourceGroups/liftie-test/providers/Microsoft.Network/virtualNetworks/liftie-dev/subnets/liftie-dev.internal.0.westus2, /subscriptions/017b8edd-1cb1-4c3e-aa2a-2f62880b1d3d/resourceGroups/liftie-test/providers/Microsoft.Network/virtualNetworks/liftie-dev/subnets/liftie-dev.internal.0.westus2'' should not have a NAT gateway associated when outbound type is SLB." Target="agentPoolProfile.vnetSubnetID"'

Before the recent change, Azure accepted the default value for outboundType as loadBalancer and allowed the AKS cluster creation.

Note that the above issue is not seen: * If the subnet(s) doesn't have a user-assigned NAT Gateway associated,
- On existing clusters, and
- On the clusters post-upgrade.

References:
- Creating AKS cluster with user-assigned NAT Gateway: https://docs.microsoft.com/en-us/azure/aks/nat-gateway#create-an-aks-cluster-with-a-user-assigned-nat-gateway
- AKS Release Note: https://github.com/Azure/AKS/commit/dfd3e6176bc979d04ef5592513cd0d2ece5146df
- Determining a possible workaround to be recommended is currently in progress.
- Microsoft Case # 2205170010003095 was created as a follow-up.
Posted May 18, 2022 - 00:36 UTC