I am using a Classic Web Service with a non-default endpoint for a Update Resource activity on the Azure Data Factory. This is the error I get:
I didn't find any info on the web and couldn't figure it out myself. This website shows an example that I used by just filling in my values for mlEndpoint, apiKey and updateRessourceEndpoint:
{
"name": "updatableScoringEndpoint2",
"properties": {
"type": "AzureML",
"typeProperties": {
"mlEndpoint": "https://ussouthcentral.services.azureml.net/workspaces/xxx/services/--scoring experiment--/jobs",
"apiKey": "endpoint2Key",
"updateResourceEndpoint": "https://management.azureml.net/workspaces/xxx/webservices/--scoring experiment--/endpoints/endpoint2"
}
}
}
There is no mention of a token that needs to be passed...
this error is basically saying the apiKey you provided is invalid to perform the update resource operation. Here is some posts for your reference: https://social.msdn.microsoft.com/Forums/azure/en-US/3bb77e37-8860-43c6-bcaa-d6ebd70617b8/retrain-predictive-web-service-programmatically-when-do-not-have-access-to-managementazuremlnet?forum=MachineLearning
Please also be noted that if you modified your linked service in ADF, remember to re-deploy the pipeline as well to reflect your change in time.