Skip to content

Improve image building dev workflow

In the dev workflow, we are currently doing:

Build RPMs -> delete instance -> build image -> deploy new instance

The latter 3 parts are all in redeploy.sh. This issue is meant to address the following:

  • Allow for trivial upgrading of the CTA instance by being able to an image with a different tag (instead of always using the dev tag)
  • Swap delete instance and build image so that the workflow follows the standards
  • Extract the image building and loading part from redeploy.sh and move this (with options) into build_image.sh

The main thing this will do is pave the way for improving the development workflow by being able to more quickly have a working CTA instance up and running (by being able to upgrade the instance, instead of tearing it down and spawning a new one).

Most likely this will need 1 follow-up issue where the plain pod definitions are converted into deployments/statefulsets, since Helm might not be able to upgrade plain pod definitions nicely