Include in Docker image necessary drivers for pyodbc

[cindyweng]

trafficstars

Can you please make it possible to write to Azure SQL Database from this Docker image using pyodbc? It is unable to be installed because of missing drivers. Thanks!

[mouhannadali]

in the conda_deoendcies.yml ; I added something like this: dependencies:

The python interpreter version.

Currently Azure ML Workbench only supports 3.5.2 and later.

• python=3.7.*

Required by azureml-defaults, installed separately through Conda to

get a prebuilt version and not require build tools for the install.

- psutil=5.6 #latest

• pyodbc=4.0.28

  Required packages for AzureML execution, history, and data preparation.

• anaconda
• pip ......

and in inference_config.yml u need to add extraDockerfileSteps: (a file include the following:) RUN apt-get -y update && apt-get -y install curl

RUN apt-get -y update &&
apt-get install -y apt-transport-https &&
curl https://packages.microsoft.com/keys/microsoft.asc | apt-key add - &&
curl https://packages.microsoft.com/config/debian/9/prod.list > /etc/apt/sources.list.d/mssql-release.list &&
apt-get update &&
ACCEPT_EULA=Y apt-get install msodbcsql17 unixodbc-dev -y

but I am getting another error now (No module named 'ruamel')

[j-so]

Apologies for the late reply!

Unfortunately, we won't be able to get to this right now. My recommendation is to use your own custom image: https://github.com/microsoft/MLOpsPython/blob/master/docs/custom_container.md

[muditmathur2020]

Hi @cindyweng, I was able to resolve this issue by creating a custom "run-config" environment for the ML-Ops Python,

We can install pyodbc and any other pip packages by going to ml_service/pipelines/diabetes_regression_build_train_pipeline.py

On line 37 and 38, we can replace the code with the one I provided below. This will create custom run-config and then manually we can enter the libraries to be used here.

##Defining the run configuration for the Pipeline to run
run_config = RunConfiguration()

# The compute on which the pipeline would be running 
# and training the model
run_config.target = aml_compute

run_config.environment.python.conda_dependencies = \
CondaDependencies.create(
conda_packages=['pandas','scikit-learn','numpy','seaborn','matplotlib',
'pyodbc','openpyxl'],pip_packages=['azureml-sdk',
'azureml-dataset-runtime[fuse,pandas]','azureml-defaults',
'azureml-interpret','python-dotenv'], 
pin_sdk_version=False)

---

Code Before:

Code After:

I hope this resolves this issue. Contributors: @muditmathur2020 and @OmkarChoulwar

Regards, Mudit Mathur Data Scientist

[OmkarChoulwar]

Hi @cindyweng, I was able to resolve this issue by creating a custom "run-config" environment for the ML-Ops Python,

We can install pyodbc and any other pip packages by going to ml_service/pipelines/diabetes_regression_build_train_pipeline.py

On line 37 and 38, we can replace the code with the one I provided below. This will create custom run-config and then manually we can enter the libraries to be used here.

##Defining the run configuration for the Pipeline to run
run_config = RunConfiguration()

# The compute on which the pipeline would be running 
# and training the model
run_config.target = aml_compute

run_config.environment.python.conda_dependencies = \
CondaDependencies.create(
conda_packages=['pandas','scikit-learn','numpy','seaborn','matplotlib',
'pyodbc','openpyxl'],pip_packages=['azureml-sdk',
'azureml-dataset-runtime[fuse,pandas]','azureml-defaults',
'azureml-interpret','python-dotenv'], 
pin_sdk_version=False)

Code Before:

Code After:

I hope this resolves this issue. Contributors: @muditmathur2020 and @OmkarChoulwar

Regards, Mudit Mathur Data Scientist

To summarize, adding any custom library in the environment is possible using the CondaDependencies in the azureml python. sdk.