Tolerations were verified to work by adding toleration (code below) which allowes scheduling of pods on master nodes of OpenShift cluster.

tolerations:
  - effect: "NoSchedule"
    key: "node-role.kubernetes.io/master"
    operator: "Exists"
    value:

Custom tolerations and node taints can be added. Unfortunately, the tolerationSeconds field is unusable. When this field is added to toleration it breaks both the reconciliation process as well as scheduling of the pods. Stateful set won't scale at all - which is based on my current understanding an OpenShift issue.