Ensuring Successful Operation of a Cloud Solution practice questions

Cloud Debugger

Why wrong: Cloud Debugger inspects application state, not endpoint availability.

Cloud Trace

Why wrong: Cloud Trace is for latency analysis, not uptime monitoring.

Cloud Logging

Why wrong: Cloud Logging is for storing and querying logs, not for uptime monitoring.

Use gcloud compute instances update --machine-type while the instance is running.

Why wrong: The instance must be stopped before changing machine type.

Detach all disks, change machine type, then reattach disks.

Why wrong: Disk detachment is not required for changing machine type.

Create a snapshot of the disk and use it to create a new instance with the desired machine type.

Why wrong: This is unnecessary; you can resize the existing instance after stopping it.

gcloud compute instances update-container and kubectl get events

Why wrong: gcloud compute instances update-container is for managed instance groups, not GKE.

kubectl edit deployment/my-app and change the image, then kubectl rollout undo if needed

Why wrong: While editing works, the question asks to monitor progress; kubectl rollout status is needed.

gcloud container clusters upgrade and kubectl get pods

Why wrong: gcloud container clusters upgrade upgrades the cluster, not the application.

An alerting policy with a log-based trigger

Why wrong: Alerting policies trigger notifications, they don't export logs.

A log-based metric

Why wrong: Log-based metrics are for creating metrics from logs, not exporting them.

An export job in BigQuery

Why wrong: BigQuery export jobs extract data from BigQuery, not import logs.

gcloud run deploy --image my-image --traffic v1=90,v2=10

Why wrong: gcloud run deploy creates a new revision, but traffic splitting uses update-traffic.

gcloud run services update --traffic v1=90,v2=10

Why wrong: The correct flag is --update-traffic or --traffic with revision names.

gcloud run revisions update v2 --traffic 10

Why wrong: Revisions cannot be updated individually for traffic.

Use bq load with an empty file to create the table, then insert data row by row.

Why wrong: Row by row insertion is extremely inefficient.

Export the original table to Cloud Storage as Avro, then load into the new table.

Why wrong: Exporting and loading is slower and incurs additional costs.

Use bq cp (copy) command.

Why wrong: bq cp copies tables, but is only available for same-location copies; however, the question asks for most efficient, but bq cp is fast. Actually, bq cp is the most efficient. But let's correct: bq cp exists. However, the correct answer should be bq query with destination table. Let me adjust.

Cloud Monitoring

Why wrong: Cloud Monitoring can show backlog metrics but not detailed per-step latency.

Cloud Profiler

Why wrong: Cloud Profiler identifies CPU/memory bottlenecks but not request latency.

Cloud Logging

Why wrong: Cloud Logging shows logs but not latency traces.

Detach the disk, create the snapshot, then reattach.

Why wrong: Detaching a running disk can cause I/O errors.

Create the snapshot while the instance is running; snapshots are always consistent.

Why wrong: Snapshots are crash-consistent, not application-consistent, which can corrupt databases.

Use gcloud compute disks snapshot without stopping; data will be consistent.

Why wrong: Incorrect; snapshots are crash-consistent only.

Dashboard

Why wrong: Dashboards visualise metrics, they don't trigger notifications.

Uptime check

Why wrong: Uptime checks monitor availability but not threshold-based alerts.

Notification channel

Why wrong: Notification channels are endpoints, they don't define conditions.

gcloud container node-pools update default-pool --autoscaling enabled

Why wrong: The flag is --enable-autoscaling, not --autoscaling enabled.

kubectl autoscale node-pool default-pool --min 1 --max 10

Why wrong: kubectl does not manage node pool autoscaling.

gcloud container clusters update my-cluster --enable-autoscaling --min-nodes 1 --max-nodes 10

Why wrong: This enables cluster-level autoscaling, not node pool level.

resource.type="compute.googleapis.com/Instance" AND instance="my-instance" AND severity="ERROR"

Why wrong: resource.type value is gce_instance, and field names are incorrect.

resource.type="gce_instance" AND resource.labels.instance_id="my-instance" AND severity="ERROR"

Why wrong: Severity can be compared with >= to include higher severities, but this is also correct. Actually, both A and D are technically correct? Wait, severity in Cloud Logging can be compared with = or >=. But the question says 'errors', so severity>=ERROR is typical. However, option D uses =, which only matches ERROR level, not CRITICAL. So A is more accurate. But let's keep A as correct.

resource.type="gce_instance" AND labels."instance_id"="my-instance" AND severity="ERROR"

Why wrong: Labels field should be resource.labels.instance_id.

gcloud run revisions update my-service --image gcr.io/my-project/my-app:v2

Why wrong: Revisions cannot be updated; use deploy.

gcloud run services update --image gcr.io/my-project/my-app:v2

Why wrong: gcloud run services update does not have --image flag; use gcloud run deploy.

kubectl set image service/my-service my-app=gcr.io/my-project/my-app:v2

Why wrong: kubectl is for GKE, not Cloud Run.

Cloud Profiler

Why wrong: Profiler identifies CPU/memory hot spots, not request latency per service.

Cloud Logging

Why wrong: Logs can help but are not the primary tools for latency analysis.

Cloud Debugger

Why wrong: Debugger inspects application state at a point in time, not latency analysis.

Cordon the node to prevent new pods

Why wrong: Cordoning is optional; drain includes it.

Use gcloud container clusters upgrade to upgrade the node

Why wrong: Upgrading is not part of draining.

An alerting policy

Why wrong: Alerts use metrics but are not required to create the metric.

A log sink

Why wrong: Log sinks export logs, not create metrics.

A notification channel

Why wrong: Notification channels are for alerting, not metrics.

Create an alerting policy with a condition

Why wrong: An alerting policy uses a metric, but you need the metric first.

Create a log sink that exports logs to BigQuery

Why wrong: A log sink exports logs, it doesn't create a metric.

Install the Logging agent on your VMs

Why wrong: The Logging agent is for sending logs, but you already have logs; you need to define a filter.

Use gcloud compute instances update --machine-type n2-standard-8 while the instance is running

Why wrong: The instance must be stopped before changing the machine type.

Delete the instance and create a new one with the desired machine type

Why wrong: This is possible but not efficient; you can change the machine type without deleting.

Use gcloud compute instances resize --machine-type n2-standard-8 without stopping

Why wrong: There is no resize command; you must use set-machine-type after stopping.

gcloud container clusters update my-cluster --enable-autoscaling --min-nodes 1 --max-nodes 10 --region us-central1

Why wrong: Autoscaling is configured on node pools, not clusters.

gcloud container clusters update my-cluster --enable-autoscaling --min-size 1 --max-size 10 --region us-central1

Why wrong: The flags are --min-nodes and --max-nodes; also applies to node pools, not clusters.

gcloud container node-pools update my-pool --cluster=my-cluster --autoscaling --min 1 --max 10 --region us-central1

Why wrong: The correct flag is --enable-autoscaling, not --autoscaling.

Create a log-based alert in Cloud Logging that triggers on network errors

Why wrong: Log-based alerts are for logs, not for uptime monitoring of external endpoints.

Use Cloud Functions to periodically call the endpoint and send an email on failure

Why wrong: This is possible but not the recommended GCP-native monitoring service.

Configure a TCP health check on the load balancer

Why wrong: Health checks are for internal load balancing, not external uptime monitoring with email notifications.

gcloud run revisions traffic my-service --to-revision new-revision --percent 10

Why wrong: The correct command is 'services update-traffic'.

gcloud run services update my-service --image gcr.io/my-project/my-image:new

Why wrong: This updates the service to use a new image but does not control traffic splitting.

gcloud run deploy my-service --image gcr.io/my-project/my-image:new --traffic new-revision=10

Why wrong: The deploy command can set traffic but only on initial deploy; for an existing service, use update-traffic.