Routing requirements

Routing requirements ensure proper connectivity between OpenShift nodes and Fusion Data Foundation storage through Multus networks.

For Fusion Data Foundation Persistent Volume Claims (PVCs) to work properly, the Ceph-CSI driver pods deployed to each OpenShift node (workers and storage) must use host networking. Ceph-CSI drivers must also be able to reach Fusion Data Foundation Ceph cluster through the Multus public network. This means that each OpenShift node must have a connection to the public network, and vice versa.

To make this as simple as possible without risk of IP overlap issues, select a different subnet (IP range) for node connections to the public network and Pod connections to the public network. Set up routing on the node public subnet and Pod public subnet such that they route to each other. The final result is that the node public subnet and Pod public subnet will act together as a single contiguous public network.

If the cluster network is used, do not route the Pod public network or node public network to the cluster network. The cluster network should be independent and isolated. The diagram below illustrates subnet routing requirements: