when setting auto_scaler_profile, new-pod-scale-up-delay gets "0s" values instead of default and autoscaler does not work as expected

[previ]

When provisioning/updating an AKS cluster, if the auto_scaler_profile block is present, then the new-pod-scale-up-delay parameter is set to 0s instead of 10s.

With this configuration the actual behaviour of the autoscaler is to wait 10 minutes (instead of proceeding immediately) from the condition of Pod "Pending", to start provisioning a new node.

Terraform (and AzureRM Provider) Version

Terraform: 0.12.29
Provider: 2.40.0

Affected Resource(s)

• azurerm_kubernetes_cluster

Terraform Configuration Files

N/A

Expected Behaviour

• The new-pod-scale-up-delay (as well as other non-terraform managed parameters) should get the default values.
• The autoscaler should work according to the parameters

Actual Behaviour

• The new-pod-scale-up-delay gets a values different from default.
• The autoscaler does not work according to the set parameter neither as the default value

Steps to Reproduce

Provision an AKS cluster with auto_scaler_profile block set.

Important Factoids

N/A

References

N/A

[stephybun]

Hi @previ,

Thanks for raising this issue and apologies for the late response. You've no doubt already seen that new_pod_scale_up_delay was added by #9291 and released in v2.44.0 of the Provider. For any other properties that cannot be set via Terraform yet I would ask you to fill out a feature request template. Thanks!

[EppO]

Not sure why this ticket has been closed and tagged as enhancement. According to the docs, new_pod_scale_up_delay default is 10s. It is not. When you manually set the parameter to 10s, value gets changed from 0s to 10s. So 0s seems to be the current default which contradicts the docs.

      ~ auto_scaler_profile {
          ~ new_pod_scale_up_delay           = "0s" -> "10s"
            # (16 unchanged attributes hidden)
        }

[stephybun]

Support for the new_pod_scale_up_delay property was only added a month after this was opened thus I interpreted the issue as an enhancement request. Thank you for the clarification though @EppO, reopening the issue.

In addition I was not able to reproduce this behaviour with v2.99.0 of the provider so in order to look into this it would be great if you could provide the following information:

• Terraform config and the steps necessary to reproduce the behaviour described above
• Terraform version with which this occurs

Thanks!