Wait for control plane pods to appear - connection refused

[alanbchristie]

We're experiencing connection refused during the OpenShift Wait for control plane pods to appear task: -

TASK [openshift_control_plane : Wait for control plane pods to appear] *************************************************************
Friday 04 October 2019  14:58:23 +0200 (0:00:00.142)       0:09:56.410 ******** 
failed: [cloudv200.zdv.Uni-Mainz.DE] (item=etcd) => {"attempts": 60, "changed": false, "item": "etcd", "msg": {"cmd": "/bin/oc get pod master-etcd-cloudv200.zdv.uni-mainz.de -o json -n kube-system", "results": [{}], "returncode": 1, "stderr": "The connection to the server cloudv200.zdv.uni-mainz.de:443 was refused - did you specify the right host or port?\n", "stdout": ""}}

[alanbchristie]

• Changing port to 8443, removing and re-installing does not help.
• Updating to 3.11.149 does not help
• Reproducing yacker base image packages does not help (on all nodes: 199, 200, 201, 202, 203)

Currently waiting for a response on the topic, in case it's a system/port issue.

Yacker does this to the base images: -

sudo yum -y update
sudo yum -y install wget git unzip net-tools yum-utils bind-utils \
      httpd-tools iptables-services ntp bridge-utils bash-completion \
      kexec-tools sos psacct pyOpenSSL etcd flannel java-1.8.0-openjdk-headless \
      docker-1.13.1 \
      atomic
sudo rpm -V docker-1.13.1
sudo systemctl enable docker
sudo systemctl start docker
sudo yum -y install NetworkManager
sudo systemctl enable NetworkManager
sudo systemctl start NetworkManager

[alanbchristie]

Appears to have been fixed by using the following patch in the inventory.root template file: -

# Following required (for some reason) for JGU installation.
# etcd starts but it listens on a specific address
# and API cannot reach it because it connects to localhost (127.0.0.1).
# See: https://github.com/openshift/openshift-ansible/issues/6986
etcd_listen_client_urls: 'https://0.0.0.0:2379'