Automatically use `.gitlab/auto-deploy-values.yaml` in Auto DevOps
Problem to solve
It's common for users to want to configure a bunch of the helm chart values used in the Auto DevOps deployment.
At the moment there are 2 ways for a user to do this:
- Using
HELM_UPGRADE_EXTRA_ARGSwith some--setfor each value you want to override. This is quite limiting in that encoding complex values is difficult in another env var - Using
HELM_UPGRADE_EXTRA_ARGSwith--valuesto point to a different YAML file in your repo as described in https://docs.gitlab.com/ee/topics/autodevops/#build-and-deployment
Option 2 is a good one since it allows you to version control the configuration which is generally encouraged in DevOps anyway but it seems a shame that this is not very obvious to users and requires a little trick that is not well documented and also requires the user to configure at least one thing as an environment variable which is not version controlled.
Intended users
Further details
Proposal
I think we should just allow users to add a YAML file to the repo and detect it automatically rather than needing to set HELM_UPGRADE_EXTRA_ARGS to point to it.
If the user adds a .gitlab/auto-deploy-values.yaml to their repo it should automatically be detected by Auto DevOps and used to override any values.yaml.
Basically we'd just add the --values .gitlab/auto-deploy-values.yaml to all the helm commands if the file is present.