Description of problem: During upgrade from v3.7 to v3.9, master-config.yaml and node-config.yaml were not backup before upgrade. This should be required for downgrade if upgrade failed. Version-Release number of the following components: openshift-ansible-3.9.12-1.git.0.6155f6e.el7.noarch How reproducible: always Steps to Reproduce: 1. Upgrade ocp from v3.7 to v3.9 2. 3. Actual results: master-config.yaml and node-config.yaml were not backup before upgrade Expected results: backup master-config&node-config file Additional info: Please attach logs from ansible-playbook with the -vvv flag
These files should be backed up each time that they're modified. Are you not finding backups generated at all?
Will release note and seek a solution in 3.9.z; potential dupe of a BZ that Fabian is working.
# ansible -i hosts masters -m shell -a "ls -la /etc/origin/master/|grep master-config" preserve-jliu-ha-master-etcd-2.0319-cwz.qe.rhcloud.com | SUCCESS | rc=0 >> -rw-r--r--. 1 root root 6085 Mar 18 22:18 master-config.yaml -rw-r--r--. 1 root root 6545 Mar 18 22:18 master-config.yaml.orig preserve-jliu-ha-master-etcd-3.0319-cwz.qe.rhcloud.com | SUCCESS | rc=0 >> -rw-r--r--. 1 root root 6085 Mar 18 22:18 master-config.yaml -rw-r--r--. 1 root root 6545 Mar 18 22:18 master-config.yaml.orig preserve-jliu-ha-master-etcd-1.0319-cwz.qe.rhcloud.com | SUCCESS | rc=0 >> -rw-r--r--. 1 root root 6085 Mar 18 22:18 master-config.yaml -rw-r--r--. 1 root root 6545 Mar 18 22:18 master-config.yaml.orig Above are all master-config files after upgrade to v3.9. # cat master-config.yaml.orig |grep defaultNodeSelector defaultNodeSelector: node-role.kubernetes.io/compute=true There is a backup named "master-config.yaml.orig", but this file is not the file before upgrade when ocp stay on v3.7. Should be overwritten by the phase two upgrade(v3.8-v3.9). So user can not downgrade to v3.7 with this file. Anyway, I think timestamp should be a better way to track any changes when happened. Just as bz1555426. Mark this one as dup. Another issue need to confirm: # ansible -i hosts nodes -m shell -a "ls -la /etc/origin/node/|grep node-config" preserve-jliu-ha-master-etcd-3.0319-cwz.qe.rhcloud.com | SUCCESS | rc=0 >> -rw-------. 1 root root 1582 Mar 18 21:11 node-config.yaml preserve-jliu-ha-master-etcd-1.0319-cwz.qe.rhcloud.com | SUCCESS | rc=0 >> -rw-------. 1 root root 1582 Mar 18 21:11 node-config.yaml preserve-jliu-ha-master-etcd-2.0319-cwz.qe.rhcloud.com | SUCCESS | rc=0 >> -rw-------. 1 root root 1582 Mar 18 21:11 node-config.yaml ... I noticed that there were not node-config backup file. If that means node-config.yaml.<timestamp> or master-config.yaml.<timestamp> did not always generate during upgrade, only happened when updated in master-config or node-config. We need to explain it in the doc, Am I right? https://docs.openshift.com/container-platform/3.9/upgrading/downgrade.html *** This bug has been marked as a duplicate of bug 1555426 ***