Deploy a Customized Python interpreter


Deploy a Customized Python interpreter

I was able to run a python script using SCRIPT_COMMAND like the following example:

My question is: can a python interpreter/distribution be deployed into the Teradata infrastructure?

I managed to copy the Python Anaconda standalone distribution in a place like /var/opt/teradata/tdtemp/uiflib/anaconda/bin/python, then the python application used that environment because the first line of the script was: #!/var/opt/teradata/tdtemp/uiflib/anaconda/bin/python instead of #!/usr/bin/python. However, is there an easy way to deploy this python environment to all the machines that configure a Terdata infrastructure? 

Teradata Employee

Re: Deploy a Customized Python interpreter


I think you might find the Orange Book "R and Python Analytics with the SCRIPT Table Operator" both interesting and useful for what you want to do. Its Section 3.3.2 is all about installing the Anaconda Python distribution on a Teradata database system >= v.15.00. See details at: in-database-r-and-python-analytics-with-the-scr...

If you are using Teradata v.15.10, an issue was discovered very recently that is not covered by the above Orange Book. Specifically, Anaconda Python uses the libcrypto library v.1.0.0, which is incompatible with the needed by Teradata >=15.10. This issue will be resolved relatively soon, but until then, one will be unable to load packages like "pandas" and "statsmodels" via Anaconda Python on Teradata 15.10.