Add inherited property support to glance
Glance currently has support for storing image metadata in single or nested key-pairs. However, there are use-cases which need more capability from this metadata storage service, which would benefit from an admin-manageable list of "inherited" properties.
One use-case is having the "configuration_
A second use-case is having a cloud-administrator want to store license-cost properties on a per-image basis (like how much to charge per hour per cpu). This property can already be protected by role with the blueprint for "api-v2-
Glance would know which property keys to inherit to child images with a "inherited" property list, stored in the glance database. A new API-extension to glance would be required for a cloud-admin to manage the "inherited" property list, which should have a similar API to to that used by whatever is proposed as part of https:/
Blueprint information
- Status:
- Complete
- Approver:
- None
- Priority:
- Undefined
- Drafter:
- None
- Direction:
- Needs approval
- Assignee:
- None
- Definition:
- Obsolete
- Series goal:
- Declined for havana
- Implementation:
- Unknown
- Milestone target:
- None
- Started by
- Completed by
- Jim Lindeman
Related branches
Related bugs
Sprints
Whiteboard
This was discussed at the Havana design summit, here's a quick summary:
For snapshots: there's a 'non_inheritabl
For uploaded images: nothing to inherit (no way to know if the uploaded image was created from a base image)
For images cloned from another region: the initial copy would preserve all appropriate metadata. (Presumably the clould provider would have the properties protected the same way in all regions, so this metadata would "stick".)
So it looks like there may be nothing to do here?
-- rosmaita
Thank you for the link. I tested with a nova.conf setting like this:
non_inheritable
and confirmed that test_prop_2 was not inherited from the parent image. So individual tenants can not currently control which properties get inherited or not, only the cloud administrator, but that is good enough for our current use-cases. So at this point, I think we can cancel out this blueprint and I'll just track and assist the "api-v2-
- lindj