Cannot update the Azure ML scoring model in the pipeline activity

Category: azure data factory


SG111273 on Fri, 14 Jul 2017 11:29:35


I'm trying to retrain a ML experiment following step by step what it is written here: 

Unfortunately the Azure ML Update Resource activity always fail with the following message:

"The service principal of your linked service is invalid to generate authentication token. Please re-deploy linked service 'updatableScoringEndpoint2' with correct servicePrincipalId, servicePrincipalKey and tenant. Error Details: 'AADSTS70002: Error validating credentials. AADSTS50012: Invalid client secret is provided. Trace ID: 41c0afd3-16ed-42c6-b96b-96c6a79d3e00 Correlation ID: 6a1582a5-fca7-4992-807b-1cd5715780d2 Timestamp: 2017-07-13 12:32:50Z"

The principal I have created is of type Web/Api and has the contributor role for the Scoring Web Service.

Both the training and scoring experiments are published using the new portal.

Any clue?

Thanks in advance.

Here below the code snippet of the linked service

    "name": "updatableScoringEndpoint2",
    "properties": {
        "hubName": "retrainingdf_hub",
        "type": "AzureML",
        "typeProperties": {
            "mlEndpoint": "",
            "apiKey": "**********",
            "updateResourceEndpoint": "",
            "servicePrincipalId": "2388d130-e86e-xxxxxxxxxxxxxxxx",
            "servicePrincipalKey": "**********",
            "tenant": "abb18969-ed9d-xxxxxxxxxxxxxxxxxxx"