Description of problem: The ordering between batch_upgrade_tasks and upgrade_tasks is not being enforced. For the N..O upgrades workflow [1] the batch_upgrade_tasks are meant to be used to execute those upgrade tasks that should be executed on one node at a time. These are also expected to be run before any of the rest of the upgrade_tasks are executed. This ordering is currently broken - there is a review out at https://review.openstack.org/#/c/448602/ to fix it. Filing this bug for more info and as we will need the fix into stable/ocata too [1] https://docs.openstack.org/developer/tripleo-docs/post_deployment/upgrade.html#upgrading-the-overcloud-to-ocata-and-beyond Version-Release number of selected component (if applicable): This is only noticeable if you are testing the batch upgrade tasks, i.e. upgrading ceph as part of the ansible upgrade steps.
Based on discussion with gfidente: Impact if not resolved: batch and non-batch tasks might be triggered simultaneously on the same node which might lead into issues during upgrade and fail the whole procedure.
On stable/ocata.
We can see that all the UpgradeBatch stacks are created/updated before the Upgrade_Step ones. Moving to verified: http://paste.openstack.org/show/607884/
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, 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/RHEA-2017:1245