NVIDIA InfiniBand Networking

NCP-AIN · 60 questions

• A municipal AI factory brings up a new InfiniBand fabric and needs subnet management high availability. What should operators provision?
• Public-health InfiniBand switches need baseline services before GPU hosts attach. What initial steps should operators complete?
• A transit ML cluster finishes InfiniBand leaf provisioning and must prove subnet manager HA. What should operators test?
• County assessor InfiniBand HA subnet managers must cooperate during provisioning. Which prerequisite should documentation include?
• A city GPU fabric replaces a failed InfiniBand switch and must keep subnet manager HA intact. What should operators do?
• An emergency-management InfiniBand rack is onboarded for civic AI. What should initial configuration include?
• Water-utility engineers finish InfiniBand management IP assignment. What should they verify next for HA?
• A school-district InfiniBand lab needs first-time switch bring-up that matches other civic sites. What should staff apply?
• A housing authority expands InfiniBand by adding switches into an existing HA subnet-manager domain. What must operators preserve?
• Courts batch-inference InfiniBand fabric finishes initial provisioning. What should operators confirm next?
• A 311 model-farm InfiniBand bring-up needs a complete initial checklist. What should that checklist cover?
• A parks computer-vision InfiniBand cluster loses its primary subnet manager during a drill. What should operators validate?
• Fire/EMS InfiniBand fabric work is queued for both management HA and later tenancy. What order should operators follow?
• A civic open-data InfiniBand pod finishes HA provisioning. What should on-call documentation include?
• City planning completes InfiniBand initial configuration with an HA subnet-manager pair. What should operators prove?
• A city GPU fabric must isolate public-health and permitting AI tenants on shared InfiniBand. What should operators configure?
• Transit and water AI jobs share InfiniBand hardware but must not communicate across agencies. What should operators assign?
• After a change window, county clinic InfiniBand nodes cannot join the intended tenant. What should operators diagnose?
• Emergency-management sensitive models run on InfiniBand HCAs that must not see every partition. What should operators do?
• A school-district InfiniBand lab wants each class project isolated on shared IB hardware. What should staff use?
• Housing authority and courts share InfiniBand switches and need tenant isolation. What should operators apply as the control plane?
• 311 and SOC AI tenants accidentally share a default InfiniBand PKey. What should operators do?
• Public-works InfiniBand expansion adds new HCA GUIDs. What PKey action should operators take?
• A water-utility InfiniBand fabric reviews how a tenant should join its partition. What should operators choose carefully?
• City planning must add a new InfiniBand PKey without breaking live tenants. What approach should operators take?
• A Fire/EMS InfiniBand security review checks partition layout. What separation should operators confirm?
• A library computer-vision InfiniBand stack claims tenant isolation. How should operators verify it?
• County assessor InfiniBand auditors ask how agencies map to partitions. What should operators maintain?
• Parks InfiniBand tenant onboarding must limit access for a new civic project. What PKey practice should staff apply?
• A municipal AI factory retires a tenant from shared InfiniBand. What should operators do?
• A public-health InfiniBand fabric sees static paths congest during model training. What should operators configure?
• Transit all-reduce traffic saturates a hot InfiniBand link while inference must stay responsive. What should operators apply?
• County assessor InfiniBand jobs improve after congested flows are redistributed. What fabric feature produced that result?
• An emergency-management InfiniBand fabric must differentiate forecast jobs from batch tenants. What should operators configure?
• A city GPU InfiniBand fabric must start a large multi-rail job. What should operators verify first?
• Water-utility InfiniBand congestion persists after partial tweaks. What combined approach should operators take?
• A school-district InfiniBand lab compares fixed routing with adaptive routing under synthetic load. When does adaptive routing help most?
• Housing-authority InfiniBand path flaps decrease after fabric tuning. What should operators keep using?
• A 311 model-farm InfiniBand SLA requires responsive interactive inference queues. What should operators map that SLA to?
• Fire/EMS InfiniBand operations need clear expectations for when paths change under load. What should the NOC document?
• Parks computer-vision InfiniBand training saturates a spine link. What should operators expect adaptive routing to do?
• A civic open-data InfiniBand pod enables QoS and must keep multi-tenant traffic fair. What tuning goal should operators meet?
• City planning changes InfiniBand subnet-manager HA and must keep traffic classes correct. What should operators re-verify?
• Public-works InfiniBand jobs fight unfairly for congested paths. What should operators select to improve multi-job fairness?
• Courts InfiniBand batch windows must not outrank daytime interactive inference. How should operators set QoS?
• A municipal AI factory expands InfiniBand racks and needs live fabric visibility. What should operators use?
• Public-health InfiniBand bandwidth drops during inference. Where should operators look first in UFM?
• Transit ML operators want warning before InfiniBand jobs fail on a dead link. What should they configure?
• A county clinic updates InfiniBand switch firmware and needs proof of impact. What UFM practice should operators follow?
• An emergency-management NOC wants one primary daily view of InfiniBand link and bandwidth health. What should they select?
• A city GPU fabric sees uneven InfiniBand utilization across rails. What should operators inspect in UFM?
• Water-utility on-call repairs an InfiniBand link and needs confirmation it is healthy again. What should they check?
• School-district staff see InfiniBand job slowdowns and need a fabric correlation habit. What should they learn to do?
• A housing-authority InfiniBand expansion may go live only after fabric health looks clean. What acceptance check should operators require?
• A 311 model-farm InfiniBand fabric needs recurring civic ops visibility. What should operators schedule?
• A Fire/EMS InfiniBand fabric sees a degraded optical link before host jobs fail. Where should operators spot it first?
• Parks computer-vision InfiniBand jobs time out while one spine port collapses in bandwidth. What should operators connect?
• A civic open-data InfiniBand pod wants fabric health in the city observability stack. What should operators do?
• City planning runs an InfiniBand maintenance window and must catch unexpected link flaps live. What should operators watch?
• Courts InfiniBand inventory was updated with new switches and HCAs. What should operators validate in UFM?