Description of problem: The FFU upgrade procedure consists of one step tagged as nova_hybrid_state in which the nova-compute container gets replaced from Queens to Train in the compute nodes. The set of steps involved in this operation are no idempotent enough and perform a restart of the docker service in the compute nodes everytime the tasks are run even though it wouldn't be required as the config files haven't been modified. We need to modify the tasks so that the docker service restart occurs only when the registries.conf has been modified, not always as it is done right now: - name: Set container_registry_insecure registries when: container_registry_insecure_registries != [] shell: crudini --set /etc/containers/registries.conf registries.insecure registries "[{{ container_registry_insecure_registries | map('regex_replace', '(.*)', "'\1'") | join(',') }}]" - name: Restart docker when: container_registry_insecure_registries != [] service: name: docker state: restarted Version-Release number of selected component (if applicable): How reproducible: Steps to Reproduce: 1. Run the FFU procedure 2. Execute the hybrid step and verify that docker service has been restarted 3. Re-run the code again and confirm that it has been restarted again Actual results: Expected results: Additional info:
16:14 <lbezdick> http://rhos-ci-logs.lab.eng.tlv2.redhat.com/logs/rcj/DFG-upgrades-ffu-ffu-upgrade-13-16.1_director-rhel-virthost-3cont_2comp-ipv4-vxlan-HA-no-ceph-from-latest_cdn/119/undercloud-0/home/stack/overcloud_upgrade_run-controller-0.log.gz 16:14 <lbezdick> Remove iscsid container before applying new paunch config 16:14 <lbezdick> 2021-05-07 17:24:08 | TASK [Remove iscsid container before applying new paunch config] *************** 16:14 <lbezdick> 2021-05-07 17:24:08 | Friday 07 May 2021 17:23:50 +0000 (0:00:07.064) 0:00:13.115 ************ 16:14 <lbezdick> 2021-05-07 17:24:08 | skipping: [controller-0] => {"changed": false, "skip_reason": "Conditional result was False"} 16:14 <lbezdick> 2021-05-07 17:24:08 | changed: [controller-1] => {"changed": true} 16:14 <lbezdick> 2021-05-07 17:24:08 | changed: [controller-2] => {"changed": true} 16:14 <lbezdick> passed ^^ ctrl0 was skipped as it's already EL8 yet the workaround fired.
Since the problem described in this bug report should be resolved in a recent advisory, it has been closed with a resolution of ERRATA. For information on the advisory (Red Hat OpenStack Platform 16.1.6 bug fix and enhancement advisory), and where to find the updated files, follow the link below. If the solution does not work for you, open a new bug report. https://access.redhat.com/errata/RHBA-2021:2097