Feature Request: 
Hi,
Openshift-marketplace operator pod should be running on the infra nodes, how to implement this using (nodeselector and tolerations)?

• Why does the customer need this? 
  OpenShift platform components are free of charge so we should be able to schedule these on infra/master nodes We want to give our end-users the full capacity of the worker nodes, if platform components are scheduled on the worker nodes, we can't guarantee them the full capacity.
•  What is the business impact, if any, if this request will not be made available? >
  We can't give our end-user the promised capacity and we will need to change our billing model. We will need a proper explanation on why platform components are consuming applications' workload resources as rRedHat clearly communicates that this shouldn't be the case.

Related Documents:

• https://access.redhat.com/solutions/6273701

Feature Goal:

• allow customers to put catalog pods on infrastructure nodes

Background:

Pod placement is usually done with node selectors and taints/tolerations, this is how ask to control pod placements of operands that supply Openshift infrastructure service, e.g. Quay. Doing this for the default catalog pods in OpenShift is harder, because there is no configuration surface for that at the CatalogSource level. The namespace openshift-marketplace is controlled by CVO and exists from Day 1, it is not possible to add namespace-level node selectors / default taints/tolerations on Day 2.

Expected deliverables:

• configuration surface on CatalogSource to add taints/tolerations and node selectors to image based catalogs to steer placement of the resulting pod on to a particular node type