PCDOE sample questions with answers
The other 180 questions are in the practice tool, with answers revealed as you go.
Start 20 questionsApplying SRE principles
- A data ingestion service has an availability objective of 99.5 percent over a 28-day window. At the end of the window, measured availability is 99.8 percent. How much of the error budget remains unspent?
- In the SRE hierarchy of service needs, which capability sits at the base, so that incident response, postmortems, and capacity planning all depend on it?
- What is the main reason a team writes an error budget policy?
- A team shipped 200 production changes last quarter. Thirty-six of those changes needed a hotfix, rollback, or patch to restore normal service. Which DORA metric does this count describe?
- A new release of a recommendation service makes responses about 40% slower. Latency still sits inside the 300 ms SLO, and no alerts have fired yet. What is the most appropriate next step for the team?
Building and implementing CI/CD pipelines
- A team sends a new build to two percent of live users, watches error rates and latency for an hour, and then raises the share in steps only if those metrics stay healthy. What is this rollout pattern called?
- A retailer needs a release with no user-visible downtime, and it wants traffic restored to the previous version within seconds if the new one misbehaves. It accepts running double capacity during the switch. Which approach fits?
- A team wants to expose a new release to a sliver of live users, watch error rates, and widen exposure only if the signals stay healthy. Which strategy matches?
- Engineers keep patching live Deployments with kubectl, and the cluster no longer matches what the repository says. Which GitOps practice addresses this?
- A team wants to release a new checkout flow to five percent of users and turn it off within minutes if errors rise, without another build. Which practice enables this?
- A build step needs an npm access token. The team must avoid storing it in the repository or in trigger definitions. What should they do?
Managing incidents
- Customers report slow checkout pages that began right after a deployment. The incident commander is already in the channel. Under standard incident response practice, which action comes first?
- A checkout service has used nearly all of its quarterly error budget, but a large marketing-driven feature is scheduled to ship next week. Under a typical error budget policy, what should the team do?
- After a corrupted-data incident, the technical trigger is known, but the team still does not understand why tests and reviews failed to catch it. Which analysis technique would best uncover that gap?
- A six-person SRE rotation is redesigning its primary on-call coverage. Which shift length best aligns with common SRE guidance for limiting burnout?
- A platform team plans an exercise in which it injects a regional database failure into a staging environment and observes whether the on-call runbook and alerts behave as expected. What is the main value of this exercise?
- A platform team moved an engineer from the networking group to own the billing service. Two weeks later, a billing incident pages that engineer at night. They have never run the billing runbooks and do not know its dependencies. What does this illustrate, and what helps?
- A platform team adds an SLO-based alert that fires when the error budget burn rate is high enough to exhaust the budget within hours, even though users have not yet reported problems. Which monitoring approach does this represent?
Optimizing performance and cost
- A Deployment has a HorizontalPodAutoscaler targeting 70 percent average CPU. Observed utilization climbs to 90 percent. What does the HPA do?
- A team wants Kubernetes to recommend or apply right-sized CPU and memory values for each container based on observed consumption, instead of changing the number of replicas. Which component fits this need?