Module 04 · CPU scheduling · Drill

CPU scheduling drill

Scenario prompts that check whether cpu scheduling is usable under pressure. Answer out loud, then compare.

4 prompts20 minNew

0 of 4 graded

Prompt 1
A shared worker fleet runs two kinds of jobs from one FIFO queue: 2-second interactive report previews that users wait for, and 40-minute bulk exports. When three exports land at once, previews wait half an hour. Redesign the scheduling, name the classic policies you are borrowing, and explain how you avoid starving exports.
Say your answer out loud first. Thirty seconds is enough.
Prompt 2
Compute average waiting time for jobs arriving at time 0 with CPU needs of 8, 4, 2 and 1 ms under FCFS in that order, under SJF, and under round robin with a 2 ms slice. Then explain which you would want for an interactive shell and why.
Say your answer out loud first. Thirty seconds is enough.
Prompt 3
A latency-sensitive Go service in Kubernetes has requests of 1 CPU and limits of 2 CPUs on 64-core nodes. Average CPU is 0.6 cores, yet p99 latency spikes to 90 ms every few seconds. Explain the mechanism and the fix, and say how you would prove it.
Say your answer out loud first. Thirty seconds is enough.
Prompt 4
In an audio-processing app, a high-urgency real-time thread fills the sound card buffer every 5 ms; a low-urgency logging thread shares a mutex with it; and a medium-urgency analytics thread is CPU-heavy. Users hear dropouts. Explain the bug and two fixes, and say what Linux scheduling classes you would use.
Say your answer out loud first. Thirty seconds is enough.