docker image does not include 'pymssql'

[ishioni]

Apache Airflow version: 1.10.10

Kubernetes version (if you are using kubernetes) (use kubectl version): 1.15.11

Environment: dockerhub image apache/airflow:1.10.10

What happened:
When importing from airflow.hooks.mssql_hook import MsSqlHook, airflow gives an error No module named 'pymssql'

What you expected to happen:
It should work :)

The docker image does not include pymssql from pip

[boring-cyborg]

Thanks for opening your first issue here! Be sure to follow the issue template!

[potiuk]

Thanks! Will take a look!

[alexbegg]

pymssql is installed with the airflow extra "mssql". It is not pre-installed with the docker image apache/airflow:1.10.10. I am currently installing it as a separate pip install of apache-airflow[mssql]==1.10.10 in my helm chart and I have no issue using the MsSqlHook.

[potiuk]

Yep. We want to minimise the prod image size to get the most frequently used packages (and core ones). We've added a new feature today to the prod image - you can add additional extras or python deps and b build your own version of the image. https://github.com/apache/airflow/blob/master/IMAGES.rst#production-images

Breeze supports building they production image and it wil even help you to add the extras in. For example:

./breeze build-image --production-image --additional-extras mssql

So I consider this solved with this feature via #8866