On all masters

1. Remove the following keys from /etc/origin/master/master-config.yaml kubernetesMasterconfig.apiServerArguments.storage-backend kubernetesMasterconfig.apiServerArguments.storage-media-type
2. systemctl stop atomic-openshift-master atomic-openshift-master-api atomic-openshift-master-controllers atomic-openshift-node

On all etcd hosts

1. systemctl stop etcd
2. mv /var/lib/etcd/member ~/bad-etcd

On first etcd host

1. cp -a /var/lib/etcd/openshift-pre-migrate-#######/member /var/lib/etcd/
2. chown -R etcd:etcd /var/lib/etcd/member
3. restorecon -RF /var/lib/etcd/member
4. echo "ETCD_FORCE_NEW_CLUSTER=true" >> /etc/etcd/etcd.conf
5. systemctl start etcd
6. etcdctl2 cluster-health
7. systemctl stop etcd
8. sed -i /ETCD_FORCE.*/g /etc/etcd/etcd.conf
9. systemctl start etcd
10. etcdctl2 member list
11. etcdctl2 member update ID####### https://ip.ip.ip.ip:2380
12. etcdctl2 member add ETCD_NAME_from_2nd_etcd_host https://ip2.ip2.ip2.ip2:2380 copy output
13. on 2nd etcd host, append to end of /etc/etcd/etcd.conf
14. on 2nd etcd host, syststemctl start etcd
15. etcdctl2 cluster-health # verify two healthy members
16. etcdctl2 member add ETCD_NAME_from_3rd_etcd_host https://ip3.ip3.ip3.ip3:2380
17. on 3rd etcd host, append onto /etc/etcd/etcd.conf
18. on 3rd etcd host, systemctl start etcd
19. etcdctl2 cluster-health # verify three healthy members

On all masters, one at a time

1. systemctl start atomic-openshift-master-api atomic-openshift-master-controllers atomic-openshift-node

Verify openshift integrity

On 2nd and 3rd etcd host

1. remove appended lines from /etc/etcd/etcd.conf
2. systemctl restart etcd
3. etcdctl2 cluster-health # verify three healthy members