M5.2 module quiz
Ten questions. 80% to pass (8 of 10). Open book, unlimited retakes.
Answers are collapsed under each question. Answer first, then check.
Q1
Across dev, test, stage and prod, the environment usually left unscheduled is:
A. Dev
B. Prod
C. Test
D. Stage
Show answer
Correct: B. Everything else is a candidate. Stage is the interesting case: it is frequently prod-like enough that teams treat it as unschedulable when it is not.
Q2
Scale-to-one on a weekend preserves:
A. Quorum maintained across the remaining nodes
B. Nothing; it is scale-to-zero
C. Full redundancy at lower cost
D. Availability of the endpoint at reduced capacity
Show answer
Correct: D. A single replica cannot hold quorum. Scale-to-minimum, at three or more, is the pattern for quorum-based systems.
Q3
Applying scale-to-one to an etcd or Kafka cluster:
A. Is the recommended cost pattern
B. Is prevented by the platform itself
C. Breaks quorum and risks the cluster
D. Has no effect on the cluster at all
Show answer
Correct: C. It is the classic way to lose a weekend, and it is why the quorum question has to be asked before the pattern is chosen rather than after.
Q4
Rolling test environments scale up per-test and down after, trading:
A. Cost for forecast accuracy
B. Cost for cold-start latency
C. Latency for reliability
D. Nothing; it is free
Show answer
Correct: B. Mitigating the cold start is what makes the pattern usable, and it is the part most implementations skip.
Q5
A freeze window is best implemented as:
A. A time-bounded override with an expiry
B. Deleting the schedule for that day
C. Pausing the whole scheduler service
D. Manual intervention on the morning
Show answer
Correct: A. The expiry is what stops a December freeze surviving into March, which is the failure that makes teams distrust freezes.
Q6
Freeze communication should be:
A. Silent, to avoid unnecessary noise
B. Per-team only, in their own channel
C. Announced in advance with the exact window
D. After the fact, in the weekly review meeting
Show answer
Correct: C. A silent freeze produces exactly the same symptoms as a broken schedule, and the resulting investigation costs more than the announcement would have.
Q7
Demo environments are frequently forgotten because:
A. They are small and cheap to run
B. They are untagged by default
C. They run in a separate account
D. Nobody owns them after the demo
Show answer
Correct: D. An environment created for one event with no owner afterwards is the archetype of un-owned spend, which is why auditing for them is a recurring exercise.
Q8
The right schedule for a sales demo environment is:
A. Business hours in the sales team’s timezone, with overrides for scheduled demos
B. Always on, to be safe
C. Weekends only, so that the team can prepare demos before the week starts
D. Never scheduled at all
Show answer
Correct: A. Overrides handle the exception. Always-on treats a rare event as a constant, which is the same error as sizing for peak.
Q9
A four-week schedule rollout typically starts with:
A. Production first, since that is where most of the spend sits
B. All environments simultaneously
C. The largest environment first
D. The environment with the clearest owner and the lowest risk
Show answer
Correct: D. The first schedule is as much about establishing trust as about savings, and a broken schedule in a high-visibility environment costs the whole programme.
Q10
Combining schedules with autoscaling requires deciding:
A. Which one wins during overlap
B. Which provider to use
C. Whether to disable autoscaling
D. Nothing; they are independent
Show answer
Correct: A. A schedule setting capacity while an autoscaler adjusts it produces oscillation unless the interaction is deliberate, which is why the two are covered together.
What’s next
Back to Schedule design patterns.