toddkitta / adventureworkstravel Goto Github PK

I was running through the lab, and stumbled when trying to consume the ML web service. The application kept throwing an exception:
"Prediction did not succeed. Check the Settings for mlWorkspaceId, mlServiceId, and mlApiKey."

Each of the settings was correct (checked, checked, and triple checked) so I downloaded the repo and debugged locally to find that the call to the ML service was continuously throwing 401 unauthorized errors.

The problem ended up being that the code is expecting the web service URI to be of this format:
"https://{0}.services.azureml.net/subscriptions/{1}/services/{2}/execute?api-version=2.0&details=true";

But the actual URI was of this format:
"https://{0}.services.azureml.net/workspaces/{1}/services/{2}/execute?api-version=2.0&details=true";

I'm not sure if this was a misstep for me somewhere along the line or not, others in the workshop did not have this issue. Either way, it might be worthwhile to consider changing the /subscriptions/ to be parameterized as well? Or maybe changing the template creation to just ask for the Request-Response URL directly and parse out the portions you need elsewhere?

edit: Just noticed that the URI actually used to be /workspaces/ until commit 827e970. Is it possible I received the old URI because I had signed up with a machine learning account prior to today?