Hey team, after multiple trials, thanks to your support, I managed to overcome my previous issues and deploy both the data processing and data science pipelines in Azure ML 😀. Now, you can see that the overall Kedro procedure has been executed successfully in Azure ML. The only paradox is that by checking the std_log.txt file, there are a few permission error messages (see the attached). Should I consider those or ignore them since they might be related to the previous errors I was facing and maintained in the memory?

Are you trying to write to “local” paths in the container during pipeline execution?

the fact is that when I don't include the code directory in the pipeline (~), the model is not able to be deployed, generating the following error

So there are two options in the plugin: a) with code upload b) with docker For a) you have to have pre-build docker image (registered as environment). In this option, every time you do

kedro azureml run

from your local machine, the code will be uploaded to Azure and executed there. For b) every time you want to run the pipeline in Azure ML you have to re-build the docker image and push it to ACR. As you can clearly see, the error is thrown from

kedro-viz

package - it seems Kedro Viz that tries to save data into local paths (local paths in Azure ML machine that runs your pipeline), which is not supported. Maybe @Nero Okwa can chip in - do you know what’s happening there? Maybe it’s some experiment tracking thing 🤔

Thanks a lot!