The open questions of the study, Q1 to Q14. The summary table lists them, and each question has a section that gives
why it matters and its status.
Note
Key answers
Open questions Q1–Q14: all are NOT AVAILABLE, except Q10 and Q14, which are answered
from the source CODE, and the control-flow part of Q6, which is also answered from the source CODE
(Questions). Q1 asks whether running VMs survive a nova_libvirt restart or recreate.
These questions cannot be settled by reading the sources, except Q10 and Q14, which are answered from the source in
their sections: Q10 by comparing the Ansible 2.8.20 and 2.9.27 code, Q14 by reading the kolla image sources. For Q6,
the Ansible 2.9.27 code answers whether later waves proceed; the state of the failed node and the availability risk
stay NOT AVAILABLE. The others are NOT AVAILABLE.
Control flow answered from the sourceCODE: a node that fails its WSREP wait marks only its own host failed, and the next wave still restarts its nodes (roles/mariadb/handlers/main.yml:66-78, roles/mariadb/tasks/restart_services.yml:28-39; ansible 2.9.27 lib/ansible/plugins/strategy/__init__.py:548-561,1019-1020,1045-1063; no MariaDB play sets any_errors_fatal; R9, fact sheet B §1.5). NOT AVAILABLE Whether the failed node is then still joining, whether that can leave a single Synced node acting as donor, and so how large the quorum and availability risk is
Answered from the sourceCODE: the code that produces F1 and F2 is the same in both versions. Loop notify: ansible 2.8.20 lib/ansible/plugins/strategy/__init__.py:540-557 matches 2.9.27 :609-626, and _ansible_notify is set the same way (2.8.20 lib/ansible/executor/task_executor.py:784, 2.9.27 :790). Tag inheritance: Task._get_parent_attribute, Block._get_parent_attribute and TaskInclude.build_parent_block (2.8.20 lib/ansible/playbook/task.py:427-458, block.py:300-364, task_include.py:124-143) match 2.9.27's task.py:458-489, block.py:300-364 and task_include.py:135-154 (Tags §2.1, §2.4)
Answered from the sourceCODE, see Kolla images §4. Nova's bootstrap containers run with KOLLA_UPGRADE: api_db sync + db sync (nova-api), db sync --local_cell (nova-conductor), so Nova's online data migrations never run outside upgrade; the images' KOLLA_BOOTSTRAP branches are unreachable. Placement's KOLLA_BOOTSTRAP runs db sync + online_data_migrations on every deploy/reconfigure. Glance adds db_load_metadefs. Heat creates heat_user_domain/heat_domain_admin only if the domain is absent, so a changed heat_domain_admin_password is never pushed to Keystone. neutron-server's KOLLA_BOOTSTRAP migrates neutron, -dynamic-routing, -fwaas and -vpnaas; the rolling upgrade list omits -dynamic-routing. RabbitMQ writes the Erlang cookie only in its bootstrap container. Agent, OVS and OVN images have no one-shot branch