Presently the Adoption instructions do not mention how should we treat the OVN Networker nodes (the OVN gateway functionality often living in Controler nodes). Should controller nodes be repurposed into Networkers? Should OVN gateways move into pods in OpenShift? Should we support both options? All scenarios that we support should be documented and ideally also tested.

There was an initial discussion at the NextGen meetup which indicates that we should always adopt the Controller/Networker nodes as Networker nodes in EDPM, to avoid a situation where the OVN gateway nodes are unmanageable – neither managed by old Director, nor new EDPM. After the original Networkers are adopted into EDPM, the admin can decide to scale them up on other nodes, and subsequently scale down the original Networkers, allowing those machines to be decomissioned if desired (assuming Ceph Mons have been moved off too). Always first managing the original OVN gateways via EDPM also allows the potential scale up + scale down + decomissioning operation to be done in a different maintenance window, reducing risk associated with the main Adoption procedure.

NextGen meetup notes: https://docs.google.com/document/d/1ddw_RmD9w6nqOblVgNw-txQ5D1SgCaApNrYzZFTrI_8/edit#heading=h.hjd4ptspql9