[update/upgrade/ffwd] no-op removal does not take effect
Affects | Status | Importance | Assigned to | Milestone | |
---|---|---|---|---|---|
tripleo |
Fix Released
|
High
|
Unassigned |
Bug Description
It seems that our approach of doing converges with --update-plan-only has no effect. The plan gets updated but the Heat stack never gets to know that, as a follow up stack operation (e.g. any `overcloud deploy`) will not use anything from the previously updated plan. The plan gets cleared and freshly re-uploaded.
This is a problem at least for parameter_defaults (e.g. resetting CephAnsiblePlaybook will never happen) but may be a problem for resource_registry too. (More investigation on resource_registry would be nice.)
The bottom line is that we probably need to do a stack update on all converge operations, and likely also do Ceph update/upgrade *before* converge, to save on the amount of stack updates we have to do.
Changed in tripleo: | |
milestone: | none → rocky-2 |
Related fix proposed to branch: stable/queens /review. openstack. org/566019
Review: https:/