Allow variable mapping in lagoon-build

[tobybellwood]

In order to better support non-lagoon based 12-factor apps, we should allow for the creation of variable maps in the build phase , so that the resultant variables are available on the lagoon-env configmap.

This step should occur prior to creating the deployment, but after all other (DBAAS) steps

Such an example might be for DB credentials - either as individual variables

DATABASE_HOST=${MARIADB_HOST:-mariadb}
DATABASE_PORT=${MARIADB_PORT:-3306}
DATABASE_NAME=${MARIADB_DATABASE:-lagoon}
DATABASE_USERNAME=${MARIADB_USERNAME:-lagoon}
DATABASE_PASSWORD=${MARIADB_PASSWORD:-lagoon}

or connstrings:

DATABASE_URL=postgres://${POSTGRES_USERNAME:-lagoon}:${POSTGRES_PASSWORD:-lagoon}@${POSTGRES_HOST:-postgres}:${POSTGRES_PORT:-5432}/${POSTGRES_DATABASE:-lagoon}

or configuring routes/hosts

WP_CONTENT_URL=${LAGOON_ROUTE}/content

etc etc

This is especially useful for images that have none of our entrypoints that would usually allow this conversion in .lagoon.env files

[shreddedbacon]

How does this work when I requested 2 mariadbs?

[shreddedbacon]

I get the idea behind this, but I'm not sure how it could be built. The user would have to inform Lagoon somehow to craft these/this new variable from provided variables.

If I requested a maria-database instead of mariadb then I'd need some way to inform the mapper to create this instead DATABASE_HOST=${MARIADB_DATABASE_HOST:-mariadb}

[tobybellwood]

I think any mapper would have to live in code in the repo (lagoon.yml 🙊?), and be managed by the developer, so it would be on them.

[shreddedbacon]

So this is currently possible using .env or .lagoon.envx files within the repository itself. Only downside is it isn't in the configmap.

Is there a specific reason it has to be in the configmap?

[tobybellwood]

If the docker image isn't uselagoon/* we have no access to the entrypoints without hacking or rebuilding the image

[shreddedbacon]

Ah yep, that makes sense. I'll have a think about this