Add project/user association to placement
This cycle we are changing the quota system to count resources to check
quota instead of tracking usage and reservations separately. As things
currently stand, we must query cell tables to count things like cores
and ram to check against quota limits. There are a couple of problems
with the current approach:
1. Querying all cells for instances owned by a project and summing their
cores and ram counts is not efficient.
2. Quota usage becomes effectively "freed" if contact with one or more
cells is lost for any reason, until the cells return.
To address these problems, we propose adding project and user associations
to placement for consumers.
Blueprint information
- Status:
- Complete
- Approver:
- Sylvain Bauza
- Priority:
- Medium
- Drafter:
- melanie witt
- Direction:
- Approved
- Assignee:
- melanie witt
- Definition:
- Approved
- Series goal:
- Accepted for pike
- Implementation:
-
Implemented
- Milestone target:
-
pike-3
- Started by
- Matt Riedemann
- Completed by
- Matt Riedemann
Related branches
Related bugs
Sprints
Whiteboard
Gerrit topic: https:/
Addressed by: https:/
Add project/user association to placement
Addressed by: https:/
Amend spec for placement project/user association
Gerrit topic: https:/
Addressed by: https:/
Add database migration and model for consumers
Addressed by: https:/
WIP Add project_id and user_id to PUT /allocations
Addressed by: https:/
WIP Add GET /usages
Addressed by: https:/
placement: use separate tables for projects/users
All changes merged as of June 24, 2017. Marking this complete for Pike. -- mriedem 20170625
Addressed by: https:/
Add missing microversion documentation
Addressed by: https:/
placement: Make API history doc more consistent