mirror of
https://github.com/jlengrand/engine.git
synced 2026-03-10 15:49:32 +00:00
60507c185d
Before, we were using kube-state-metrics from Prometheus operator chart. But because we can't control resource usage, it may be preferable to use the dedicated chart for it. I disable it because Terraform was failing on chart upgrade (because of pod CrashloopBackoff)
Naming convention
- providers: providers files should be named tf-providers-
- variables: should be named with _, never with - (in the tf-default-vars.j2.tf)
- variables should start with the name of the service associated (ex: eks_, rds_...)
- tf filenames: filenames should never contain _ or they may not be interpreted
- resources names should not contain - but _ instead (resource "type" "name" {})
- name field in resource should not contain _ and - instead to avoid RFC 1123 limitation (resource "type" "name" {name = "this name"})
- the length of fields names should never exceed 64 chars and prefer less than 32 chars, still for RFC 1123
- any and only *.j2.tf files will be rendered by jinja