Search code examples
pythonconfigurationcontinuous-integrationconfiguration-filescontinuous-deployment

Managing configurations for different environments

I've had a discussion with some people at work and we couldn't come to a conclusion.
We've faced a dilemma - how do you manage different configuration values for different environments?

We've come up with some options, but none of them seemed to satisfy us:
- Separate config files (i.e. config.test, config.prod etc.), and having a file pointing to the selected one (at ~/env for instance), or an environment variable pointing to it.
- Using a single DB to store all configurations (you query it with your environment and get the corresponding configuration values)
- Creating configuration files on deploy (Using CI/CD system like Atlassian Bamboo)

Which is the more widely used option? Is there a better way?
Should config file be kept in the git repository along with the rest of the code? Our systems are written in python (both 2.7 and 3)

Solution

  • We ended up using a method similar to this one. We had a base settings file, and environment specific files that simply imported everything from the base file base.py:

    SAMPLE_CONFIG_VARIABLE = SAMPLE_CONFIG_VALUE

    prod.py: 

    from base.py import *

    So all we had to do when accessing values from the configuration is read an environment variable and creating the file corresponding to it.