Azure ML kernel

Question

Azure ML kernel

Marko Cvokic 1

Hi,

I installed custom env and added it to jupyter on my compute instance.
Yet, when I run the code error says that it occurs under base env (azureml_py38)

see the pic. Can you help me?

romungi-MSFT 49,096 Microsoft Employee Moderator

@Marko Cvokic In this case with the error message it looks like the default kernel is used. Guessing by the kernel name chosen I believe you want to use python 3.7 and I have tried this scenario with my script and env.yml and the environment seems to use the custom environment. These are the scripts and yaml files used to create compute.

env.yml  
    name: myenv  
    channels:  
        - conda-forge  
    dependencies:  
        - python==3.7  
        - pip:  
            - arrow  
            - ipykernel

Script.sh

#!/bin/bash  
   
set -e  
  
# This script creates a custom conda environment and kernel based on a sample yml file.  
  
conda env create -f env.yml  
echo "Activating new conda environment"  
conda activate myenv  
conda install -y ipykernel  
echo "Installing kernel"  
sudo -u azureuser -i <<'EOF'  
conda activate myenv  
python -m ipykernel install --user --name myenv --display-name "mykernel"  
echo "Conda environment setup successfully."  
EOF

Maybe, you could try again or select a different kernel and then try with your custom environment to check if it loads correctly?

Your answer

romungi-MSFT 49,096 Reputation points Microsoft Employee Moderator

2021-11-17T13:11:03.82+00:00

@Marko Cvokic In this case with the error message it looks like the default kernel is used. Guessing by the kernel name chosen I believe you want to use python 3.7 and I have tried this scenario with my script and env.yml and the environment seems to use the custom environment. These are the scripts and yaml files used to create compute.

env.yml name: myenv channels: - conda-forge dependencies: - python==3.7 - pip: - arrow - ipykernel

Script.sh

#!/bin/bash set -e # This script creates a custom conda environment and kernel based on a sample yml file. conda env create -f env.yml echo "Activating new conda environment" conda activate myenv conda install -y ipykernel echo "Installing kernel" sudo -u azureuser -i <<'EOF' conda activate myenv python -m ipykernel install --user --name myenv --display-name "mykernel" echo "Conda environment setup successfully." EOF

Maybe, you could try again or select a different kernel and then try with your custom environment to check if it loads correctly?

Share via

Azure ML kernel

Your answer