Anyscale Kubernetes, Slurm & GPU Scheduling interview questions

Choose the control plane from job admission, placement, recovery and service needs, with a concrete partial-allocation failure.

See how partial allocation can strand two training jobs. Separate group admission, real placement, readiness and recovery instead of assuming atomic pod startup.

Balance durable checkpoint cost against replay and restart loss, with explicit assumptions and a warning sign when the approximation breaks down.

Build a durable, bounded scheduling loop with reservations and checkpoint-aware preemption, and avoid universal queue-wait formulas.

Trace Kubernetes pod placement and Ray bundle reservation as separate layers, including logical GPU limits and autoscaling delay.

Choose the tenant boundary before planning base-model reuse. Account for frozen weights, adapter state and activations separately, then schedule by compatibility, fairness and the real cost of restarting training.

Calculate traffic across the actual inter-cloud boundary before deciding. Whole-job placement often avoids repeated transfers, but communication volume, overlap, convergence requirements and available links determine whether a split job is viable.

Trace the job from quota reservation through admission to physical placement. Count eligible node shapes, not just free GPUs, and use controller and scheduler evidence to choose the least disruptive fix.

Separate quota, borrowing, reclaim latency and fairness, with a demand-conserving example of when pooling improves utilization.

Follow a four-GPU request from advertised capacity to concrete devices, and see why free GPUs across nodes cannot satisfy one pod.

DRA gives the scheduler device claims and attributes to work with. Compare that with extended-resource counts, then check the deployed driver and controller support before promising sharing, partitioning or gang placement.

Diagnose GPU pod failures by scheduling, container startup and execution boundaries, without treating Allocatable as a health certificate.

Separate discovery labels, GPU resource advertisement and validation, and understand the operator as a reconciler with configurable components.

Use actual rank placement and measured job duration to decide whether waiting for locality beats starting immediately.

Define warm, cached and truly cold paths, then size transfer budgets and warm capacity without promising universal sub-second restores.

Compare the resource promises of MIG, time-slicing and MPS, then test peak memory and simultaneous demand before packing workloads.

Design fractional GPU admission around explicit shapes and enforceable guarantees, then reconcile reservations with the runtime.

Make readiness mean that the engine can serve, scale from measured demand and ready capacity, and drain streams before termination. Kubernetes can implement these controls; an operator packages and reconciles them.

Size usable inference headroom from the complete readiness delay and demand ramp, then decide which training capacity can safely be borrowed.

Choose a sandbox from its threat model and supported workload, then calculate density from total session memory and CPU demand. GPU access needs its own isolation and compatibility review.

Measure scheduling, image transfer, unpacking and model initialization separately. A missing-device error needs its own checks for allocation, container access and CUDA compatibility.

Compare the cost of committed capacity with the demand each additional GPU would actually serve. Keep billing discounts separate from capacity guarantees, and account for bursts without counting the same work twice.

Count feasible placements for the waiting job, including its per-node and topology requirements. Then compare waiting, reservations and checkpoint/restart consolidation using their actual disruption costs.