Pass network requests to scheduler
Since network requests can be useful for scheduler to make network related placement decisions, it makes sense to pass this piece of information into scheduler, so that proper destination hosts can be selected before instances being built and networks being allocated.
Currently, the network request is passed to nova from the ‘create’ method of compute api on instance booting requests. It goes all the way through to the conductor, which takes this as a param called ‘requested_
nova.scheduler.
nova.scheduler.
nova.compute.
In this process, the ’requested_
This blueprint addresses this issue and aims to pass ‘requested_
The changes involved in the scheduler utils and conductor manager would Not depend on, or affect the functionality of request_spec object currently being implemented, and no api changes will be required. However we do recommend that the future request_spec object take ‘requested_
Blueprint information
- Status:
- Started
- Approver:
- None
- Priority:
- Undefined
- Drafter:
- Xinyuan Huang
- Direction:
- Needs approval
- Assignee:
- Xinyuan Huang
- Definition:
- New
- Series goal:
- None
- Implementation:
- Started
- Milestone target:
- None
- Started by
- Xinyuan Huang
- Completed by
Related branches
Related bugs
Sprints
Whiteboard
Gerrit topic: https:/
Addressed by: https:/
Proposes passing network requests to scheduler
Gerrit topic: https:/
Addressed by: https:/
Pass requested_networks to scheduler
Work Items
Work items:
Modify nova.scheduler.
Modify the future RequestSpec object by adding requested_networks as a field: TODO