1. Outline the proposed title of this feature request.

Propagate nodeSelector/tolerations configuration on the operator subscriptions

1. What is the nature and problem description of the request?

Currently it is possible to configure nodeSelector/tolerations for the ODF operator itself in its Subscription CR:

apiVersion: operators.coreos.com/v1alpha1
kind: Subscription
metadata:
  name: odf-operator
  namespace: openshift-storage
spec:
  channel: stable-4.20
  config:
    nodeSelector:
      cluster.ocs.openshift.io/openshift-storage: ''
  name: odf-operator
  source: redhat-operators
  sourceNamespace: openshift-marketplace 

However such configuration is not propagated automatically on the children operators subscriptions:

apiVersion: operators.coreos.com/v1alpha1
kind: Subscription
metadata:
  name: cephcsi-operator-stable-4.20-redhat-operators-openshift-marketplace
  labels:
    operators.coreos.com/cephcsi-operator.openshift-storage: ''
spec:
  channel: stable-4.20
  config:
    tolerations:
      - effect: NoSchedule
        key: node.ocs.openshift.io/storage
        operator: Equal
        value: 'true'
  name: cephcsi-operator
  source: redhat-operators
  sourceNamespace: openshift-marketplace

As a result we are compelled to manually patch the children operators subscriptions CR one by one.

1. Why does the customer need this? (List the business requirements here)

Many AI customers have only tainted worker GPU nodes to prevent the scheduling of non AI business related workload in those expensive nodes. They don't have any other worker node types. They expect this ODF children operators to run in infra node in order to:

• save subscriptions from pure platform related workload
• save space in their expensives GPU nodes 

1. Are there any Documentation Requirements for this request?

No

1. Is the request coming from a specific customer or subset of customers (on prem only, cloud only, etc). Please don't mention any customer related sensitive information.

Coming from AI customers mainly running OCP on premise

1. What is the urgency of the request?

Not urgent but very high importance since it can make the product installation non possible on constraint environment having only infra + tainted GPU nodes.

1. Please provide contact information, in case of follow up questions.
   Laurent TOURREAU: laurent.tourreau@redhat.com