We should add a CI step that lints the committed kubernetes resources, and errors if there are pods without resource requests. We'll need to handle at least two separates modes of this, which would be us directly committing the kubernetes yaml, as well as our use of helm charts, which would indirectly generate the kubernetes yaml. We should also have a way to ignore this linter in certain spots, to account for extenuating circumstances.
We should add a CI step that lints the committed kubernetes resources, and errors if there are pods without resource requests. We'll need to handle at least two separates modes of this, which would be us directly committing the kubernetes yaml, as well as our use of helm charts, which would indirectly generate the kubernetes yaml. We should also have a way to ignore this linter in certain spots, to account for extenuating circumstances.