Pyspark environment and python packages on executors

[rbarata]

At my company we use conda-pack to make certain packages available on the spark executors. Is there a better a better alternative to get away from creating a new environment and pack it every time I need a new python lib to be available for the executors?

Cheers 🙂

[-werners- · Accepted solution]

databricks provides library installation in the form of PyPi packages, or wheel/egg.

If you install the packages like that on the cluster, they are automatically sent to all executors.

[Kaniz_Fatma]

Hi @Ricardo Barata​, We haven’t heard from you on the last response from @Werner Stinckens​ , and I was checking back to see if his suggestions helped you. Or else, If you have any solution, please do share that with the community as it can be helpful to others.