I'm following a tutorial to learn how to use Azure ML Studio. Yesterday, I completed the Use Azure Machine Learning Designer without a hitch and got a diabetes prediction service up and running (I completed the first two parts of the tutorial before that, also without problem).
Today, I wanted to re-do the steps to really get them in my head, but at the Run the training pipeline step, hitting Submit no longer displayed the progress for each component (yesterday, I could see each component going from yellow to green, and observe outputs once green - all of that in the pipeline canvas of the Designer - as in the example image below - before a Submit, the green band isn't there).
What I did
Checked that my compute instance was running (OK)
Checked that my Complute Cluster was running (seemed ok, but noticed there was no jobs history, and that my pipeline build, when checking general Jobs, only took 4 seconds (instead of the usual 5-15min) - so I deleted and rebuilt it, as it seemed dead.
The new
Submitoperation took the expected ~10 minutes, but still no visual of the progress showing in the Pipeline Canvas.Went to check the jobs finished within the cluster itself, and saw that everything ran ok (I could see the outputs here)
I couldn't find any logs, nor got any error message telling me something was wrong.
Here are my questions:
- Am I missing something obvious (entirely possible since I'm new to Azure ML)
- Is there a place to really monitor what's going on under the hood of Azure ML Studio to double check the validity of a "Successful" job
- Has anyone else experienced this and found a way to "fix" it (so that I can see the training on the canvas, per component, as it progresses) ?
Thank you !
