Skip to content
This repository has been archived by the owner on Aug 2, 2023. It is now read-only.

Install kabanero operator at the cluster scope #701

Open
kaczyns opened this issue May 11, 2020 · 0 comments
Open

Install kabanero operator at the cluster scope #701

kaczyns opened this issue May 11, 2020 · 0 comments

Comments

@kaczyns
Copy link
Member

kaczyns commented May 11, 2020

We are transitioning to allow multiple Kabanero instances in the cluster. Rather than install the Kabanero operator in each namespace, we'll have the operator installed at the cluster scope. This is similar to how other operators (ie Serverless, Tekton Pipelines, etc) are installed.

There is an operator-sdk page that describes what is necessary from the operator perspective, when transitioning to cluster scope. It mainly revolves around changing the watch namespace, and converting the roles and bindings to their cluster cousin names:
https://sdk.operatorframework.io/docs/operator-scope/

Since we install using OLM, all of these changes will be done in the CSV. I think that we already give the kabanero operator cluster-admin authority so it's likely that no changes are required here.

The installModes in the CSV should change to only support AllNamespaces.

I am guessing that OLM will not let you upgrade from a single namespace CSV to a all namespaces CSV, however it's a moot point since a cluster scoped operator should get installed in the openshift-operators namespace. The install script should detect that the 0.9.0 version is installed in the kabanero namespace, remove it, and then install this version in the openshift-operators namespace.

We should continue to assume for the time being that the Kabanero CR instance will be created in the kabanero namespace, although that assumption will change soon.

Sign up for free to subscribe to this conversation on GitHub. Already have an account? Sign in.
Labels
None yet
Projects
None yet
Development

No branches or pull requests

1 participant