Ceph-Ansible upgrade fails when RHCS RGW nodes are in a separate subnet to the remaining nodes in the cluster

Issue

• In a Red Hat Ceph Storage 4 cluster with the RGW nodes in a subnet that is different to the remaining nodes in the Ceph cluster, Ceph-Ansible upgrades will fail with the error The task includes an option with an undefined variable. The error was: No first item, sequence was empty on task set_fact _radosgw_address to radosgw_address_block ipv4. An example of the Ceph-Ansible log file showing the error is shown below.

  <TIMESTAMP> p=<PID> u=<USER> n=ansible | TASK [ceph-facts : set_fact _radosgw_address to radosgw_address_block ipv4] ********************************************************************************
  **************************************************************************************************
  <TIMESTAMP> p=<PID> u=<USER> n=ansible | fatal: [<IP_ADDRESS>]: FAILED! => 
    msg: |-
      The task includes an option with an undefined variable. The error was: No first item, sequence was empty.
  
      The error appears to be in '/usr/share/ceph-ansible/roles/ceph-facts/tasks/set_radosgw_address.yml': line 11, column 3, but may
      be elsewhere in the file depending on the exact syntax problem.
  
      The offending line appears to be:
  
      - name: set_fact _radosgw_address to radosgw_address_block ipv4
        ^ here