Description of problem:
When run upgrade_control_plane.yml on atomic hosts, the atomic-openshift-node and openvswitch image weren't upgraded.
Version-Release number of selected component (if applicable):
Steps to Reproduce:
1. install OCP 3.4 HA Env on Atomic Host
2. run v3_5/upgrade_control_plane.yml
3. check pods status
4. run v3_5/upgrade_node.yml
5. check pods status
The atomic-openshifit-node and openvswitch use v3.4 images on master hosts
The atomic-openshifit-node and openvswitch should be upgraded by upgrade_control_plane.yml
Jan, Did we somehow not include containerized node upgrades in the work you did to include node upgrades on masters during control plane?
Scott, Jan the issue also exist with RPM installation. I think the reason is that only partial PR were merged into atomic-openshift-utils-3.5.4-1.git.0.034b615.el7.noarch
Try with atomic-openshift-utils-3.5.6-1.git.0.5e6099d.el7.noarch, all PR was merged into it. But the issue still exist.
that could be related to the inventory file. Can you share its content? Currently, only masters that are nodes at the same time are upgraded:
- name: Drain and upgrade master nodes
For RPM installation, The atomic-openshift-node wasn't restarted in master host. You can check the pid by 'systemctl status atomic-openshift-node'
For containerized installation, The atomic-openshift-node is still use v3.4. you can use 'docker ps' command.
Resolved by https://github.com/openshift/openshift-ansible/pull/3380
Merged. Anping Li, can you verify?
Jan, the fix works. you can build a puddle for further more testing.
Scott, is building a puddle something you can do?
I have, but for some reason the installer isn't making it into the errata puddle. It's in the non errata puddle though. I'll sort this out today.
The nodes was upgraded by upgrade_control_plane.yml, so move bug to verified.