NCP-MCI-6.10 · Question #40
In a five-node cluster, an administrator noticed that three VMs are consuming too many resources on a single host. Acropolis Dynamic Scheduling (ADS) is not able to migrate these VMs. What is the…
The correct answer is B. VMs use GPU pass-through. Acropolis Dynamic Scheduling (ADS) performs automatic workload balancing by live-migrating VMs between hosts. However, VMs that use GPU pass-through (direct hardware assignment/passthrough of a physical GPU) cannot be live migrated. This is because GPU pass-through creates a…
Question
In a five-node cluster, an administrator noticed that three VMs are consuming too many resources on a single host. Acropolis Dynamic Scheduling (ADS) is not able to migrate these VMs. What is the most likely reason preventing ADS from migrating these VMs?
Options
- AVMs use a Volume Group.
- BVMs use GPU pass-through.
- CVM-VM anti-affinity policy is set.
- DVMs use external Network Attached Storage.
How the community answered
(43 responses)- A7% (3)
- B79% (34)
- C2% (1)
- D12% (5)
Explanation
Acropolis Dynamic Scheduling (ADS) performs automatic workload balancing by live-migrating VMs between hosts. However, VMs that use GPU pass-through (direct hardware assignment/passthrough of a physical GPU) cannot be live migrated. This is because GPU pass-through creates a direct, exclusive hardware dependency between the VM and the specific physical host's GPU - the virtualization layer cannot intercept and transfer that state to another host. ADS detects this condition and excludes these VMs from migration candidates. VMs using Volume Groups or external NAS can generally be migrated. VM-VM anti-affinity would only prevent co-location, not restrict migration away from an overloaded host.
Topics
Community Discussion
No community discussion yet for this question.