region=infra is not the tag our playbooks take into account. The nodes should have node-role.kubernetes.io/infra=true label before upgrade to avoid being marked as compute nodes. 

This, however, fails with node-role.kubernetes.io/infra=true too, as there is no group associated with this node, so it would get compute label during bootstrapping.

There are several options how to solve this:
1) Guess the openshift_node_group from existing labels
2) Ensure openshift_node_group is set
3) ??

Scott, should we enforce openshift_node_group during upgrade?

Yes, we've said that we are going to force all hosts to have openshift_node_group set for them. We need to document this too.

We'll check to ensure that openshift_node_group_name is set for every single host so that there's no longer any ambiguity as to how node groups are being assigned to hosts.

Need to document this requirement.

Hi, Vadim.
I observed that now after upgrade to v3.10, the node labels are different from the one with openshift-ansible-3.10.0-0.27.0.

one infra node has one role (infra), 
the other node has two roles (infra & compute)

[root@wmengug39r75-master-etcd-zone1-1 ~]# oc version
oc v3.10.0-0.53.0
kubernetes v1.10.0+b81c8f8
features: Basic-Auth GSSAPI Kerberos SPNEGO

Server https://104.197.218.186
openshift v3.10.0-0.53.0
kubernetes v1.10.0+b81c8f8
[root@wmengug39r75-master-etcd-zone1-1 ~]# oc get nodes
NAME                                STATUS    ROLES           AGE       VERSION
wmengug39r75-master-etcd-zone1-1    Ready     master          1h        v1.10.0+b81c8f8
wmengug39r75-master-etcd-zone2-1    Ready     master          1h        v1.10.0+b81c8f8
wmengug39r75-master-etcd-zone2-2    Ready     master          1h        v1.10.0+b81c8f8
wmengug39r75-node-zone1-primary-1   Ready     compute         1h        v1.10.0+b81c8f8
wmengug39r75-node-zone2-primary-1   Ready     compute         1h        v1.10.0+b81c8f8
wmengug39r75-nrriz-1                Ready     infra           1h        v1.10.0+b81c8f8
wmengug39r75-nrriz-2                Ready     compute,infra   1h        v1.10.0+b81c8f8
[root@wmengug39r75-master-etcd-zone1-1 ~]# oc get nodes --show-labels
NAME                                STATUS    ROLES           AGE       VERSION           LABELS
wmengug39r75-master-etcd-zone1-1    Ready     master          1h        v1.10.0+b81c8f8   beta.kubernetes.io/arch=amd64,beta.kubernetes.io/instance-type=n1-standard-4,beta.kubernetes.io/os=linux,failure-domain.beta.kubernetes.io/region=us-central1,failure-domain.beta.kubernetes.io/zone=us-central1-a,kubernetes.io/hostname=wmengug39r75-master-etcd-zone1-1,node-role.kubernetes.io/master=true,role=node
wmengug39r75-master-etcd-zone2-1    Ready     master          1h        v1.10.0+b81c8f8   beta.kubernetes.io/arch=amd64,beta.kubernetes.io/instance-type=n1-standard-4,beta.kubernetes.io/os=linux,failure-domain.beta.kubernetes.io/region=us-central1,failure-domain.beta.kubernetes.io/zone=us-central1-b,kubernetes.io/hostname=wmengug39r75-master-etcd-zone2-1,node-role.kubernetes.io/master=true,role=node
wmengug39r75-master-etcd-zone2-2    Ready     master          1h        v1.10.0+b81c8f8   beta.kubernetes.io/arch=amd64,beta.kubernetes.io/instance-type=n1-standard-4,beta.kubernetes.io/os=linux,failure-domain.beta.kubernetes.io/region=us-central1,failure-domain.beta.kubernetes.io/zone=us-central1-b,kubernetes.io/hostname=wmengug39r75-master-etcd-zone2-2,node-role.kubernetes.io/master=true,role=node
wmengug39r75-node-zone1-primary-1   Ready     compute         1h        v1.10.0+b81c8f8   beta.kubernetes.io/arch=amd64,beta.kubernetes.io/instance-type=n1-standard-4,beta.kubernetes.io/os=linux,failure-domain.beta.kubernetes.io/region=us-central1,failure-domain.beta.kubernetes.io/zone=us-central1-a,kubernetes.io/hostname=wmengug39r75-node-zone1-primary-1,node-role.kubernetes.io/compute=true,region=primary,role=node
wmengug39r75-node-zone2-primary-1   Ready     compute         1h        v1.10.0+b81c8f8   beta.kubernetes.io/arch=amd64,beta.kubernetes.io/instance-type=n1-standard-4,beta.kubernetes.io/os=linux,failure-domain.beta.kubernetes.io/region=us-central1,failure-domain.beta.kubernetes.io/zone=us-central1-b,kubernetes.io/hostname=wmengug39r75-node-zone2-primary-1,node-role.kubernetes.io/compute=true,region=primary,role=node
wmengug39r75-nrriz-1                Ready     infra           1h        v1.10.0+b81c8f8   beta.kubernetes.io/arch=amd64,beta.kubernetes.io/instance-type=n1-standard-4,beta.kubernetes.io/os=linux,failure-domain.beta.kubernetes.io/region=us-central1,failure-domain.beta.kubernetes.io/zone=us-central1-a,kubernetes.io/hostname=wmengug39r75-nrriz-1,node-role.kubernetes.io/infra=true,region=infra,registry=enabled,role=node,router=enabled
wmengug39r75-nrriz-2                Ready     compute,infra   1h        v1.10.0+b81c8f8   beta.kubernetes.io/arch=amd64,beta.kubernetes.io/instance-type=n1-standard-4,beta.kubernetes.io/os=linux,failure-domain.beta.kubernetes.io/region=us-central1,failure-domain.beta.kubernetes.io/zone=us-central1-b,kubernetes.io/hostname=wmengug39r75-nrriz-2,node-role.kubernetes.io/compute=true,node-role.kubernetes.io/infra=true,region=infra,registry=enabled,role=node,router=enabled
[root@wmengug39r75-master-etcd-zone1-1 ~]#

>I observed that now after upgrade to v3.10, the node labels are different from >the one with openshift-ansible-3.10.0-0.27.0.

>one infra node has one role (infra), 
>the other node has two roles (infra & compute)

Right, verified that the upgrade is working correctly if

   openshift_node_labels="{'region': 'infra'}"

is replaced by 

   openshift_node_group_name='node-config-infra'

Moving this to documentation team

OCP 3.6-3.10 is no longer on full support [1]. Marking CLOSED DEFERRED. If you have a customer case with a support exception or have reproduced on 3.11+, please reopen and include those details. When reopening, please set the Target Release to the appropriate version where needed.

[1]: https://access.redhat.com/support/policy/updates/openshift