Skip to content
This repository has been archived by the owner on Jan 28, 2023. It is now read-only.

Latest commit

 

History

History
50 lines (41 loc) · 3.31 KB

carbon_config.md

File metadata and controls

50 lines (41 loc) · 3.31 KB
Raw

Carbon config

Carbon config allows configuring some aspects of the Carbon behavior. Currently, Carbon config supports only one parameter (besides apiVersion):

Parameter Required Default value Description
apiVersion true version of the config format. The latest version is v1alpha1
carbonScope false cluster this parameter defines the scope of the Carbon metadata. Allowed values are: cluster and namespace.

Carbon config can be defined in multiple places:

  1. carbon-data namespace
  2. your active namespace

In the future, we're going to add the third option - locally on a workstation.

When Carbon config is defined at Kubernetes (currently it's a single option), it has to be defined at a ConfigMap with name carbon-config under the data key config. It can be defined using JSON or YAML format. This is an example of a correct ConfigMap with the Carbon config:

apiVersion: v1
kind: ConfigMap
metadata:
  name: carbon-config
data:
  config: |
    apiVersion: v1alpha1
    carbonScope: namespace

As soon as Carbon config has only one parameter, we didn't implement merging of parameters from different sources. But it's what we're going to do in the future as well.

Carbon scope

Carbon scope controlls different aspects of a package installation. Currently it supports two modes:

  • cluster
  • namespace

'cluster' Carbon scope

  • This scope is used by default.
  • Metadata for installed packages is stored at the carbon-data namespace. More details about pagckages metadata can be found at Installed package metadata.
  • If a resource has predefined namespace, this resource will be intalled to the corresponding namespace. It can be any existing namespace.
  • If a resource doesn't have a configured namespace, it will be installed to the namespace managed by --namespace command line argument (default by default).
  • You are not allowed to install multiple copies of the same Carbon package to the same Kubernetes cluster.

'namespace' Carbon scope

  • Metadata for installed packages is stored at the same namespace where the package is installed. More details about pagckages metadata can be found at Installed package metadata.
  • If a resource has predefined namespace and this NS is different from the one, managed by --namespace command line argument (default by default), the predefined NS will be overriten and a warning message will be exposed.
  • If a resource doesn't have a configured namespace, it will be installed to the namespace managed by --namespace command line argument (default by default).
  • Carbon applies additional carbon/component-namespace lable for all resources in order to avoid conflicts when Cluster-scoped resources are used.
  • You are alloed to install multiple copies of the same Carbon package to different namespaces. But it's your responsibility to provide uniq names for cluster-scoped resources.

Limitations

  • Carbon doesn't delete resources from a namespace when none resource in the NS is applied. It's a limitation of kubectl. In the future we're going to handle cleaning-up process by ourselves and it such will resolve the issue.
  • It's impossible to migrate from one Carbon scope to another at the moment.