PCDOE Practice Test 12 — 15 Questions

Question 1

Your team runs a service on GKE and wants to implement SLO-based alerting. The SLO is 99.9% availability measured over a 30-day rolling window via a request-based SLI. You configure a burn-rate alert with a 1-hour window and a burn rate threshold of 14.4. During a partial outage where the service is unavailable for 5 minutes in every hour, the alert does not fire. What is the most likely reason?

Choose an answer to begin — your selection is scored in the full session.

Accepted Answer

The SLI is computed as a ratio of good requests to total requests, and the 5-minute outages may not produce enough bad requests to exceed the burn rate threshold in the 1-hour window.. The alert does not fire because the burn rate over the 1-hour window remains below 14.4. With a request-based SLI, short outages that affect a small percentage of requests may not consume error budget fast enough to exceed the threshold. The threshold itself is appropriate for fast-burn detection; the issue is the mismatch between the outage pattern and the alert condition. To catch such patterns, consider a multi-window alert or a lower threshold for slower burns.

Answer

The burn rate threshold of 14.4 is too high; it should be set to 1.0 to detect any outage.

Answer

The alerting policy is misconfigured because it lacks a notification channel; without a notification channel, the alert never fires.

Answer

The rolling window is 30 days, but the burn rate alert should use a 30-day window to match the SLO; using 1 hour causes the alert to miss short outages.

Question 2

You are the SRE for a critical API service on Google Cloud. The service has an SLO of 99.95% availability. During a recent incident, the service was unavailable for 10 minutes. After the incident, you need to calculate the error budget consumed. Assuming a 30-day month, how much of the error budget was consumed?

Accepted Answer

About 46.3%. The error budget is the allowed downtime derived from the SLO. For 99.95% availability over 30 days, the budget is 21.6 minutes. The incident consumed 10 minutes, which is 10/21.6 ≈ 46.3% of the budget. This calculation is essential for tracking error budget consumption and making decisions about releases and reliability work.

Answer

About 23.1%

Answer

About 100%

Answer

About 2.8%

Question 3

A DevOps engineer is trying to create a service account key for a CI/CD pipeline, but receives the error: 'Constraint constraints/iam.disableServiceAccountKeyCreation violated'. What is the most likely cause and solution?

Accepted Answer

The Organization Policy prevents key creation; the engineer needs to request an exception from the security team.. The error 'Constraint constraints/iam.disableServiceAccountKeyCreation violated' indicates that an Organization Policy with the constraint `iam.disableServiceAccountKeyCreation` is enforced at the organization, folder, or project level. This policy explicitly blocks the creation of service account keys, which is why the engineer cannot generate a key for the CI/CD pipeline. The correct solution is to request an exception from the security team, who can either remove the policy or add the engineer's project to an exemption list.

Answer

The Cloud Resource Manager API is disabled; enable it.

Answer

The project does not have billing enabled; enable billing.

Answer

The service account has been deleted; the engineer must recreate it.

Question 4

You are implementing a canary deployment for a new version of a microservice on Google Kubernetes Engine (GKE). You want to automatically roll back if the canary's error rate exceeds the baseline by 2% for 5 minutes. Which approach best achieves this?

Accepted Answer

Use Cloud Monitoring to create an alert on the canary's error rate, and configure the alert to trigger a Cloud Function that rolls back the deployment.. Automated canary analysis requires monitoring error rates and triggering rollback when thresholds are breached. Cloud Monitoring can compare canary and baseline metrics, and with alerting to a Cloud Function, you can automate rollback. Kubernetes probes and Istio alone don't provide this logic. Manual rollback is not automatic.

Answer

Use Istio to route traffic to the canary, and rely on its built-in automatic rollback feature when errors exceed a threshold.

Answer

Use kubectl rollout undo to manually revert if errors are observed.

Answer

Configure a Kubernetes readiness probe that fails when errors occur, causing the canary pods to be restarted.

Question 5

Your team manages a high-traffic e-commerce API on Google Cloud. The SRE team wants to proactively detect performance regressions before they impact users. They decide to implement a canary deployment strategy. Which of the following best describes how to use Cloud Monitoring to automatically halt the canary rollout if the error rate exceeds a threshold?

Accepted Answer

Use Cloud Deploy with a canary deployment strategy and configure a deployment verification job that queries a Cloud Monitoring metric for error rate and fails the rollout if the threshold is exceeded.. Cloud Deploy provides native canary deployment support with verification jobs that can query Cloud Monitoring metrics. When the verification job detects that the error rate exceeds the defined threshold, it fails, which automatically stops the rollout and prevents further exposure. This integrates directly with the deployment pipeline, enabling safe, automated canary analysis without custom scripting or indirect triggers.

Answer

Define an SLO with a rolling window and use Cloud Monitoring alerting policies to trigger a Cloud Build pipeline that rolls back the canary when the error rate exceeds the threshold.

Answer

Create a log-based metric for HTTP 5xx errors and configure a log sink to Pub/Sub that triggers a Cloud Function to roll back the deployment.

Answer

Configure a Kubernetes Horizontal Pod Autoscaler to scale down the canary deployment if the error rate increases, effectively removing the canary from service.

Question 6

You are an SRE for a service running on Google Cloud. You want to define an SLO for a gRPC API that uses Cloud Endpoints. The service must respond to 99% of requests within 300 ms over a 28-day window. Which SLI definition should you use to measure latency?

Accepted Answer

The proportion of requests with a response time less than or equal to 300 ms, measured at the load balancer.. The correct SLI is the proportion of requests with latency less than or equal to 300 ms, measured at the load balancer. This directly matches the SLO statement of 99% of requests within 300 ms. It is a request-based SLI that captures the user experience and can be computed from access logs or monitoring metrics. Other options either use averages, percentiles, or counts, which do not directly measure the proportion of good requests as required by the SLO.

Answer

The average response time of all requests over the 28-day window, compared to 300 ms.

Answer

The 99th percentile of response time over the 28-day window, compared to 300 ms.

Answer

The number of requests that exceed 300 ms in the 28-day window.

Question 7

An SRE team wants to reduce alert noise by ensuring alerts are actionable and routed to the correct on-call rotation. They are using Cloud Monitoring with alerting policies. Which configuration should they implement?

Accepted Answer

Create alerting policies with conditions based on symptoms, such as high error rate or latency, and use notification channels that target the appropriate on-call team via email and SMS.. Symptom-based alerts, such as those monitoring error rates or latency, directly reflect user experience and are actionable. Routing them to the correct on-call team through appropriate notification channels ensures rapid response. The other options either generate excessive noise, use unreliable triggers, or suppress alerts in ways that risk missing critical incidents.

Answer

Create alerting policies for every possible metric deviation, such as CPU usage above 60%, and send all alerts to a single distribution list.

Answer

Create alerting policies that trigger only when multiple metrics exceed thresholds simultaneously, and disable notifications during business hours to reduce interruptions.

Answer

Create alerting policies based on log entries containing the word 'error' and use a webhook to create tickets in a ticketing system for all alerts.

Question 8

A DevOps engineer is reviewing GCP costs and notices high network egress charges. Which two actions can help reduce network egress costs? (Choose two.)

Accepted Answer

Use Cloud CDN to cache content closer to users. Option C is correct because Cloud CDN caches content at Google's globally distributed edge points of presence, so repeated requests are served from an edge location near the user instead of traversing the network from the origin, which directly reduces egress traffic and its associated charges. Option D is correct because keeping communicating resources in the same region means traffic between them stays on Google's internal network and is billed at lower intra-region rates (or is free for same-zone traffic), avoiding costly inter-region egress. Option A is wrong because preemptible VMs reduce compute costs, not network egress charges. Option B is wrong because consolidating workloads into a single zone does not by itself reduce egress and can harm availability; egress is driven by where traffic exits to users or crosses regions, not by zone count. Option E is wrong because external global load balancers distribute traffic globally and can increase egress by routing responses across regions rather than reducing egress costs.

Answer

Use preemptible VMs

Answer

Migrate all workloads to a single zone

Answer

Use external load balancers with global access

Question 9

Your team is adopting SRE practices and wants to implement a robust incident management process. Which two practices should you adopt to ensure effective incident response and learning? (Choose two.)

Accepted Answer

Conduct blameless postmortems after every incident to identify root causes and preventive actions.. The two practices that ensure effective incident response and learning are defining clear incident command roles and conducting blameless postmortems. Clear roles streamline coordination and communication during high-pressure situations, while blameless postmortems uncover systemic causes and drive preventive actions. Together, they form a foundation for reliable incident management and continuous improvement, as advocated by Google SRE.

Answer

Implement strict change management with multiple approval layers to prevent unauthorized changes.

Answer

Avoid documenting incident details to prevent information overload and focus on resolution.

Answer

Ensure that only senior engineers are on-call to handle incidents, as they have the most experience.

Question 10

During organization bootstrap, the platform team wants every new project created under the 'prod' folder to automatically receive a baseline set of IAM bindings and a specific VPC network peering. They cannot rely on developers to run setup scripts. What should they implement?

Accepted Answer

Implement an organization-level Cloud Asset Inventory feed or a project-creation event trigger that invokes a Cloud Function or Cloud Run job to apply the baseline configuration.. Automatically applying a baseline to projects the platform team does not create requires an event-driven mechanism. Cloud Asset Inventory real-time feeds or Logging-routed project-creation audit events deliver notifications to Pub/Sub, where a Cloud Function or Cloud Run job applies the baseline IAM and peering. This works regardless of whether projects are created by Terraform, the console, or gcloud, which is exactly what developers' self-service creation demands.

Answer

Create a Terraform module and instruct developers to reference it in their project creation code.

Answer

Enable the Compute Engine API at the organization level so new projects inherit the VPC peering automatically.

Answer

Configure an organization policy constraint that enforces the required IAM bindings on all projects.

Question 11

You are an SRE for a high-traffic e-commerce platform running on Google Kubernetes Engine (GKE). The platform experiences a sudden surge in traffic during a flash sale. To maintain reliability, you must implement an SRE practice that automatically adjusts resources based on demand while preserving the error budget. Which approach best meets these requirements?

Accepted Answer

Configure Horizontal Pod Autoscaler (HPA) with custom metrics based on request latency and CPU utilization, and set appropriate min and max replicas.. To handle sudden traffic surges while preserving the error budget, the platform needs automatic, demand-driven scaling. Horizontal Pod Autoscaler with custom metrics such as request latency and CPU utilization enables responsive scaling that aligns with actual user experience. Setting min and max replicas ensures availability and cost control. This approach maintains reliability by preventing resource exhaustion and keeping latency within SLO thresholds.

Answer

Manually increase the number of replicas in the deployment to a fixed higher number before the sale begins.

Answer

Use Vertical Pod Autoscaler (VPA) to automatically increase the CPU and memory requests of existing pods during the surge.

Answer

Set up a Cluster Autoscaler to add nodes when pods are pending, and rely on the default HPA configuration based on CPU only.

Question 12

You are an SRE for a service that uses a 99.9% availability SLO over a 30-day window. The service has consumed 100% of its error budget with 10 days remaining in the window. According to Google SRE best practices, what should you do?

Accepted Answer

Halt all changes that could impact reliability and focus on reliability improvements, while allowing low-risk changes with approvals.. The correct action is to halt risky changes and focus on reliability improvements, while permitting low-risk changes that enhance reliability. This aligns with the error budget policy: when the budget is exhausted, the team must prioritize reliability over new features. It is not a complete freeze; critical fixes and reliability enhancements can proceed with oversight. This approach protects users and allows the team to recover budget in the next window.

Answer

Increase the SLO target to 99.95% to allow more error budget.

Answer

Freeze all changes to the service until the next SLO window begins.

Answer

Ignore the error budget exhaustion because there are still 10 days left in the window.

Question 13

A company runs a stateless batch processing workload on a managed instance group (MIG) with autoscaling based on average CPU utilization. Jobs finish in about 10 minutes and are not sensitive to interruption. During month-end peaks, the MIG scales out to 120 instances and cost spikes. You want to reduce cost without lengthening the average job completion time. What should you do?

Accepted Answer

Configure the MIG to use Spot VMs with a fallback to standard VMs, and keep CPU-based autoscaling.. Spot VMs are designed for fault-tolerant, stateless, or batch workloads that can tolerate preemption. Because the jobs are short and stateless, preemption has minimal impact on average completion time, and the discount directly reduces the cost of peak scale-out. Committed use discounts, metric changes, and instance resizing do not address the spiky, interruptible nature of this workload.

Answer

Change the autoscaling metric to average memory utilization and set a lower target.

Answer

Purchase a 1-year committed use discount for the peak vCPU and memory, and keep the same autoscaling policy.

Answer

Reduce the maximum number of instances in the MIG and increase the per-instance machine size proportionally.

Question 14

You are setting up a Cloud Build trigger that should run only when changes are pushed to the `main` branch of a GitHub repository. Which configuration should you use?

Accepted Answer

Create a trigger with the `--branch-pattern` set to `^main$`.. To restrict a Cloud Build trigger to a specific branch, use the `--branch-pattern` flag with a regular expression that matches the branch name. For the `main` branch, `^main$` ensures that only pushes to `main` trigger the build. This is the standard and efficient method.

Answer

Create a trigger with the `--branch-pattern` set to `.*` and add a build step that checks the branch name.

Answer

Create a trigger with the `--tag-pattern` set to `^main$`.

Answer

Create a trigger that listens to all pushes and use a `if` condition in the build config to check the branch.

Question 15

A company runs a microservices application on Google Kubernetes Engine (GKE). They observe that many pods request more CPU and memory than they actually use, leading to low node utilization and high cost. They want to optimize resource usage without impacting application performance. Which approach should they take?

Accepted Answer

Enable the Vertical Pod Autoscaler (VPA) in recommendation mode and use its recommendations to manually adjust resource requests.. Over-provisioned resource requests lead to reserved but unused capacity. VPA in recommendation mode provides data-driven suggestions to right-size requests without automatic changes, allowing the company to adjust safely. HPA and node auto-provisioning address scaling, not request sizing, and setting limits equal to requests does not reduce waste.

Answer

Configure a Horizontal Pod Autoscaler (HPA) for each deployment based on CPU utilization.

Answer

Use node auto-provisioning to automatically add nodes when pods are pending due to insufficient resources.

Answer

Set resource limits equal to resource requests for all pods to ensure guaranteed QoS.