Fail gracefully if resources are overcommitted on a compute node
Registered by
Lorin Hochstein
If a compute node gets scheduled with a request, and it turns out that the compute node did not have the resources to handle this request (even though the scheduler thought that it did), then it should fail gracefully.
Also need some way of specifying what the maximum allowable virtual machine cpus/core/storage that can be allocated on the node before it starts to reject requests.
Blueprint information
- Status:
- Complete
- Approver:
- Vish Ishaya
- Priority:
- Undefined
- Drafter:
- Nova Orchestration Team
- Direction:
- Approved
- Assignee:
- Nova Orchestration Team
- Definition:
- Approved
- Series goal:
- None
- Implementation:
- Implemented
- Milestone target:
- None
- Started by
- Yaguang Tang
- Completed by
- Yaguang Tang
Related branches
Related bugs
Sprints
Whiteboard
Not sure if this belongs in orchestration or zones. Feel free to switch it if it seems appropriate. --Vish
(?)