AzureStack Could Not Find Member 'enableCrossZoneUpgrade' and 'prioritizeUnhealthyInstances'

[TheOnlyWei]

Describe the bug
We are getting bad request error running Azure CLI version 2019-03-01-hybrid in an Azure Stack environment. The following is the error:

az : ERROR: Deployment failed. Correlation ID: e436a4a0-c388-4834-aecf-ada3f8cf5c56. {
At C:\test\Workloads\DeveloperExperience\CLI\Workloads.Tests.CLI.ps1:759 char:23
+ ...   $result = az vmss create -g $vmssRgName -n $vmssName --admin-userna ...
+                 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
    + CategoryInfo          : NotSpecified: (ERROR: Deployme...ada3f8cf5c56. {:String) [], RemoteException
    + FullyQualifiedErrorId : NativeCommandError

  "error": {
    "details": [
      {
        "message": "Could not find member 'enableCrossZoneUpgrade' on object of type 'RollingUpgradePolicy'. Path 
'properties.upgradePolicy.rollingUpgradePolicy.enableCrossZoneUpgrade', line 1, position 338.",
        "target": "vmss.properties.upgradePolicy.rollingUpgradePolicy.enableCrossZoneUpgrade"
      },
      {
        "message": "Could not find member 'prioritizeUnhealthyInstances' on object of type 'RollingUpgradePolicy'. Path 
'properties.upgradePolicy.rollingUpgradePolicy.prioritizeUnhealthyInstances', line 1, position 374.",
        "target": "vmss.properties.upgradePolicy.rollingUpgradePolicy.prioritizeUnhealthyInstances"
      }
    ],
    "code": "BadRequest",
    "message": "The request message is invalid."
  }
} 

To Reproduce

1. Install Azure CLI.
2. Register:

            az cloud register --name $TntAdminEnvironmentName `
                 --endpoint-resource-manager $asEndpoints.ResourceManagerEndpoint `
                 --suffix-storage-endpoint $asEndpoints.StorageEndpointSuffix `
                 --suffix-keyvault-dns ("." + $asEndpoints.AzureKeyVaultDnsSuffix) `
                 --endpoint-vm-image-alias-doc https://raw.githubusercontent.com/Azure/azure-rest-api-specs/master/arm-compute/quickstart-templates/aliases.json `
                 --profile 2019-03-01-hybrid

3. Run the following command:

az vmss create -g $vmssRgName -n $vmssName --admin-username $vmssAdminUserName --admin-password $vmssAdminPassword --image $imageName --authentication-type password --upgrade-policy-mode $upgradeMode

Expected behavior
Should not throw bad request error.

Environment summary
We are using the nightly builds:

https://aka.ms/InstallAzureCliWindowsEdge

Additional Context
I have an internal email open with Feiyue Yu (@qwordy). We can provide additional information in the email or here if requested.

[yonzhan]

compute

[qwordy]

This issue was introduced in this PR. I raised my concern of this issue during PR review.

[xfz11]

Hi Wei Shi (@TheOnlyWei), This error happened because your request's api version is old.
If you feel it urgent to fix the problem, the workaround is to use az cloud set -n <cloud name> --profile latest to update your CLI cloud profile.
If you specify the profile to be 2019-03-01-hybrid, then you have to wait for it to be fix in next CLI release.

[qwordy]

Xing Zhou (@zhoxing-ms) We need to update following code. Add if statement of each parameter to check whether it is None.

azure-cli/src/azure-cli/azure/cli/command_modules/vm/_template_builder.py

Lines 932 to 939 in 716b10f

	'rollingUpgradePolicy': {
	'maxBatchInstancePercent': max_batch_instance_percent,
	'maxUnhealthyInstancePercent': max_unhealthy_instance_percent,
	'maxUnhealthyUpgradedInstancePercent': max_unhealthy_upgraded_instance_percent,
	'pauseTimeBetweenBatches': pause_time_between_batches,
	'enableCrossZoneUpgrade': enable_cross_zone_upgrade,
	'prioritizeUnhealthyInstances': prioritize_unhealthy_instances
	}

[zhoxing-ms]

Xing Zhou (@zhoxing-ms) We need to update following code. Add if statement of each parameter to check whether it is None.

azure-cli/src/azure-cli/azure/cli/command_modules/vm/_template_builder.py

Lines 932 to 939 in 716b10f

	'rollingUpgradePolicy': {
	'maxBatchInstancePercent': max_batch_instance_percent,
	'maxUnhealthyInstancePercent': max_unhealthy_instance_percent,
	'maxUnhealthyUpgradedInstancePercent': max_unhealthy_upgraded_instance_percent,
	'pauseTimeBetweenBatches': pause_time_between_batches,
	'enableCrossZoneUpgrade': enable_cross_zone_upgrade,
	'prioritizeUnhealthyInstances': prioritize_unhealthy_instances
	}

Feiyue Yu (@qwordy) Yes, that's what I thought~