Description of problem: Upgrade against rpm ocp when set "openshift_node_bootstrap=false" in hosts file, which means user want to keep node run as original way. After upgrade finished, checked that node was not run as original way but still run as a bootstrap node. before upgrade: # ps aux|grep "start node" root 21861 1.6 2.3 579760 86248 ? Ssl Apr26 5:46 /usr/bin/openshift start node --config=/etc/origin/node/node-config.yaml --loglevel=5 root 65062 0.0 0.0 112708 980 pts/0 S+ 03:45 0:00 grep --color=auto openshift after upgrade: # ps aux|grep "node" root 74544 2.8 2.6 845668 97816 ? Ssl 03:42 1:00 /usr/bin/hyperkube kubelet --v=5 --address=0.0.0.0 --allow-privileged=true --anonymous-auth=true --authentication-token-webhook=true --authentication-token-webhook-cache-ttl=5m --authorization-mode=Webhook ........ root 77029 0.2 1.7 673756 62260 ? Ssl 03:43 0:04 openshift start network --config=/etc/origin/node/node-config.yaml --kubeconfig=/tmp/kubeconfig --loglevel=2 root 77120 0.0 0.0 11684 1492 ? Ss 03:43 0:00 /bin/bash -c #!/bin/bash set -euo pipefail ..... # systemctl status openvswitch Unit openvswitch.service could not be found. # oc get pod -n openshift-sdn NAME READY STATUS RESTARTS AGE ovs-2jxrj 1/1 Running 0 25m ovs-mx74p 1/1 Running 2 25m sdn-4nxkd 1/1 Running 0 25m sdn-m96w9 1/1 Running 2 25m # docker images|grep ose-node registry.reg-aws.openshift.com:443/openshift3/ose-node v3.10 30eab79c6a92 3 hours ago 1.61 GB Version-Release number of the following components: openshift-ansible-3.10.0-0.30.0.git.0.4f02952.el7.noarch How reproducible: always Steps to Reproduce: 1. Rpm install ocp v3.9 2. Set "openshift_node_bootstrap=false" in hosts file 3. Run upgrade against above ocp Actual results: Node was run as a bootstrap node. Expected results: Node should keep a pet-node the same with it before upgrade. Additional info: Need a check about "openshift_node_bootstrap" when upgrade nodes.
openshift_node_bootstrap=false is not going to be supported, we're going to force everyone to use node bootstrapping and during the upgrade assert that they have created node configmaps for their node groups prior to the upgrade.