No schedulable nodes found matching node selector for Cluster Monitoring Operator - 'node-role.kubernetes.io/infra=true'

[vorburger]

With #4 "solved" (worked around...), this is up next:

TASK [openshift_control_plane : Ensure that Cluster Monitoring Operator has nodes to run on] ********************************************************************************
Monday 10 December 2018  18:56:05 +0000 (0:00:00.504)       0:09:07.201 ******* 
fatal: [192.168.0.11]: FAILED! => {
    "assertion": false, 
    "changed": false, 
    "evaluated_to": false, 
    "msg": "No schedulable nodes found matching node selector for Cluster Monitoring Operator - 'node-role.kubernetes.io/infra=true'"
}

Failure summary:

  1. Hosts:    192.168.0.11
     Play:     Configure Cluster Monitoring Operator
     Task:     Ensure that Cluster Monitoring Operator has nodes to run on
     Message:  No schedulable nodes found matching node selector for Cluster Monitoring Operator - 'node-role.kubernetes.io/infra=true'

[vorburger]

This is because at least one of the nodes needs to be a openshift_node_group_name='node-config-infra' instead of a openshift_node_group_name='node-config-compute', and 0f1e613 fixed this for me.

PS: At first I tought an infra node is special and won't host normal workload pod containers later, and I was looking for a way to mark a node to be BOTH a compute and an infra for a moment. But seeing compute,infra for the 2nd node in this output makes me think this is probably not so and it's fine:

[centos@openshift-master ~]$ oc get nodes
NAME                        STATUS    ROLES           AGE       VERSION
openshift-master.rdocloud   Ready     master          2d        v1.11.0+d4cacc0
openshift-node1.rdocloud    Ready     compute         2d        v1.11.0+d4cacc0
openshift-node2.rdocloud    Ready     compute,infra   2d        v1.11.0+d4cacc0