Currently, if a user wants to change a small aspect of an init container (e.g. setting custom resource limits), they have to provide the full definition of the init container in the podTemplate.
Presumably we do this to give users full control over how init containers are executed. However, there are a couple of UX issues worth considering:
- Most users are probably only interested in minor tweaks such as changing the resource limits of the container
- The way the operator treats
initContainers of the podTemplate is confusing. (Other sections in the podTemplate get merged while initContainers simply get replaced if the container name matches a built-in container)
Currently, if a user wants to change a small aspect of an init container (e.g. setting custom resource limits), they have to provide the full definition of the init container in the
podTemplate.Presumably we do this to give users full control over how init containers are executed. However, there are a couple of UX issues worth considering:
initContainersof thepodTemplateis confusing. (Other sections in thepodTemplateget merged whileinitContainerssimply get replaced if the container name matches a built-in container)