GCP Associate Cloud Engineer Practice Question

Your project hosts a managed instance group named mig-prod containing 10 Compute Engine VMs. Today, an alerting policy triggers a separate incident for every VM whose CPU usage exceeds 80 percent, flooding the on-call channel. You need a single incident only when the average CPU utilization of the whole group stays above 80 percent for at least five minutes. Which configuration accomplishes this goal?

Enable Cloud Monitoring autoscaling recommendation alerts for mig-prod so a single incident opens automatically when average CPU exceeds 80 percent.
Create an alerting policy on gce_instance/cpu/utilization with no cross-series reducer and group-by the instance_id label; trigger when any time series is above 0.8 for 5 minutes.
Configure an uptime check for one VM in mig-prod and set an alerting policy to fire when the check fails for 5 minutes; duplicate the policy for the remaining VMs.
Create an alerting policy on metric gce_instance/cpu/utilization filtered to mig-prod, use an alignment period of 60 s with per-series aligner ALIGN_MEAN, apply cross-series reducer REDUCE_MEAN with no group-by labels, and trigger when the value is above 0.8 for 5 minutes.

Report Issue

Answer Description

To alert on the fleet instead of on each VM you must first align the individual time-series and then reduce them into one combined series.
Filtering the metric compute.googleapis.com/instance/cpu/utilization (alias gce_instance/cpu/utilization) to the MIG and setting ALIGN_MEAN with a 60-second alignment period produces one data point per VM per minute.
Applying the cross-series reducer REDUCE_MEAN and leaving group-by empty collapses those per-VM points into a single average value across all 10 VMs.
Setting the condition threshold to "above 0.8" with a 5-minute duration means the policy opens only one incident after the average CPU stays above 80 % for five consecutive minutes.
The other options either evaluate each VM separately, rely on an uptime check that measures availability rather than CPU, or use autoscaling recommendation alerts which are not intended for real-time incident notification.

Ask Bash

Bash is our AI bot, trained to help you pass your exam. AI Generated Content may display inaccurate information, always double-check anything important.