HA Cluster configuration question

Hello!
Is it possible to configure cluster in way, that when failed node come back, containers, that were on that node and relocate to other during fencing, would come back to their original node?
I.e.
1) Normal functionality:
node0[ctid100, ctid103], node1[ctid102], node2[ctid104, ctid105]
2) node2 fail:
node0[ctid100, ctid103, ctid105], node1[ctid102, ctid104], node2[fenced]
3) node2 came back:
node0[ctid100, ctid103], node1[ctid102], node2[ctid104, ctid105]

Is it possible?
Thank you in advance!