This is a task about to discuss and spec out how changes to ores should be made to make it work with the blubber/helm/kubernetes. In order to achieve this, we need to archive ores deploy repo. That means:
- All of model files need to be either pulled when building the images (That would make the images gigantic) or make service pulls them down when it's starting (make the configs accept URLs as model path)
- Configs need to be overhauled into one big yaml file that's going to be checked out into helm charts
- Model files wouldn't work without the codes (features and feature lists). This means the repos need to be turned to libraries and they would be installed building the image using a file like requirements-production.txt
- Assets ores use, they also need to pulled down when starting the service.
- Scap configs need to turn to helm configs
- wheels repo need to be archived and completely ditched
- helm chart for ores need to have redis subchart for tests, meaning redis should also have a chart in releases.wikimedia.org/charts
- Speaking of tests, it's better to wrap all of python tests into one tox.ini file so entrypoint for the "test" stage of docker images can be easily defined
- The blubber files need to be added to ores repo but it's blocked on T210267: Execution of the deployment pipeline should be configurable via .pipeline/config.yaml
(More?)