PCA Practice Test 19 — 15 Questions

Question 1

Which Alertmanager feature would you use to prevent a 'ServiceDown' alert from firing if a 'GlobalNetworkOutage' alert is already active?

Accepted Answer

Inhibition. Inhibition rules allow one alert to suppress another based on matching labels.

Answer

Grouping

Answer

Silencing

Answer

Routing

Question 2

When performing vector matching between two vectors of different cardinalities, which THREE vector matching modifiers can be used? (Choose three)

Accepted Answer

on. In PromQL, when performing binary operations between two instant vectors that have different label sets or cardinalities, the on modifier restricts matching to an explicitly specified set of labels, which is essential for one-to-one matching across differing label dimensions. The group_left modifier allows a many-to-one or one-to-many match by permitting the left vector to have multiple series matching a single series on the right, which is exactly how different cardinalities are reconciled. The ignoring modifier excludes specified labels from the matching process, enabling matches between vectors whose label sets differ by the ignored labels. group_by is not a valid PromQL vector matching modifier (it is not a keyword in binary operations), and match_any is not a PromQL modifier at all, so neither belongs here.

Answer

group_by

Answer

match_any

Question 3

When configuring Alertmanager 'group_by' settings, what happens if you include a label that is not present on an incoming alert?

Accepted Answer

The alert is grouped under a null value. Alertmanager's group_by directive groups alerts by the values of the specified labels. If a label listed in group_by is absent from an incoming alert, Alertmanager treats its value as empty/null and still groups the alert under that null value rather than dropping or erroring. This means alerts missing the label are grouped together in a single notification group.

Answer

The alert is dropped

Answer

The alert triggers an error in the logs

Answer

The routing rule fails

Question 4

Which THREE of the following metrics are automatically added to all scraped targets?

Accepted Answer

up. Option B, up, is correct because Prometheus automatically synthesizes this metric for every scrape target, reporting 1 if the target was reachable and scraped successfully and 0 otherwise. Option D, scrape_samples_scraped, is correct because Prometheus automatically records the number of samples exposed by the target during each scrape. Option E, scrape_duration_seconds, is correct because Prometheus automatically measures and exposes the duration of each scrape in seconds. Options A and C, cpu_usage and memory_usage, are not automatically added by Prometheus; such host or application metrics must be exposed by the target itself (for example, via node_exporter or a client library) and are not synthesized by the scrape process.

Answer

cpu_usage

Answer

memory_usage

Question 5

Which TWO are true about Prometheus alerting rules?

Accepted Answer

They use PromQL expressions for evaluation.. Option A is correct because Prometheus alerting rules are defined with PromQL expressions in the rule file, and the configured expression is evaluated against time-series data to determine whether an alert should fire. Option B is correct because the Prometheus server itself evaluates alerting rules as part of its rule evaluation cycle, then sends firing alerts to Alertmanager for notification handling. Option C is incorrect because alerting rules only produce alert states such as pending or firing; they do not trigger a system restart. Option D is incorrect because exporters expose metrics for Prometheus to scrape, and alerts are not sent to exporters. Option E is incorrect because Prometheus stores its own time-series data locally and does not require a separate database for alerting rules.

Answer

They always trigger a system restart.

Answer

They are sent to the exporter.

Answer

They require a separate database.

Question 6

You have a global Alertmanager configuration where you want to suppress 'DiskSpaceLow' alerts if 'InstanceDown' is active for the same host. Which mechanism accomplishes this?

Accepted Answer

Inhibition rules. Inhibition rules in Alertmanager are designed to suppress a set of alerts when another set of alerts is already firing. They match on labels, so you can define an inhibition rule that mutes 'DiskSpaceLow' alerts if an 'InstanceDown' alert with the same host label is active. This is exactly the use case described: suppressing dependent alerts when a more fundamental alert is firing.

Answer

Route tree

Answer

Repeat intervals

Answer

Grouping

Question 7

Where do you define the threshold for a Prometheus alerting rule?

Accepted Answer

In the 'expr' field. In Prometheus alerting rules, the 'expr' field contains the PromQL expression that is evaluated on each evaluation interval; when the expression returns a non-empty result set, the alert transitions to pending and then firing. The threshold itself (e.g., '> 0.9' or '== 0') is embedded directly inside that PromQL expression, so the threshold is defined in 'expr'. The 'labels' field only attaches metadata (severity, team) to the alert, and Alertmanager only handles routing, grouping, silencing, and notification after Prometheus fires the alert.

Answer

In the 'labels' field

Answer

In the Alertmanager config

Answer

In the Grafana panel

Question 8

Which TWO of the following are primary components of the Prometheus monitoring architecture?

Accepted Answer

Alertmanager. The Prometheus Server (D) is the core component of the Prometheus architecture: it scrapes metrics from instrumented targets over HTTP, stores them in its local time-series database (TSDB), and evaluates PromQL recording and alerting rules. Alertmanager (B) is the other primary component, receiving alerts fired by the Prometheus Server and handling deduplication, grouping, silencing, inhibition, and routing them to receivers such as email, Slack, or PagerDuty. Fluentd (A) is a log collection and forwarding tool, not part of Prometheus's metric-based monitoring pipeline. Zipkin (C) is a distributed tracing system, and Kibana (E) is a visualization front end for Elasticsearch logs, neither of which are Prometheus components.

Answer

Fluentd

Answer

Zipkin

Answer

Kibana

Question 9

Which of the following describes the 'pull' model's benefit regarding service discovery?

Accepted Answer

Targets don't need to know where Prometheus is.. Prometheus can dynamically discover targets (via Kubernetes API, Consul, etc.) and automatically start scraping, removing the need to register every new pod.

Answer

Prometheus is always the first to start.

Answer

Targets push data when they are ready.

Answer

It reduces network traffic.

Question 10

You are monitoring a Linux server and need to track CPU usage. Which exporter is the standard choice for exposing hardware and OS metrics?

Accepted Answer

node_exporter. node_exporter is the standard Prometheus exporter for hardware and OS metrics exposed by *NIX kernels.

Answer

mysqld_exporter

Answer

blackbox_exporter

Answer

pushgateway

Question 11

You want to ensure that if the Alertmanager cluster loses communication, alerts are still sent. Which configuration helps achieve this?

Accepted Answer

Use the --cluster.peer flag. High Availability (HA) for Alertmanager is achieved by running multiple instances and using the '--cluster.peer' flag to link them.

Answer

Disable grouping

Answer

Set repeat_interval to 0

Answer

Use a load balancer only

Question 12

When configuring Service Discovery in Kubernetes, why might you use relabeling?

Accepted Answer

To reduce cardinality by dropping high-variance labels.. Relabeling in Prometheus service discovery lets you rewrite the label set of discovered targets before scraping, and a common use is dropping or aggregating high-variance labels (e.g., pod UID, ephemeral instance IDs) that would explode cardinality. Since every unique label combination creates a new time series, dropping noisy labels keeps the TSDB manageable and queries fast. This is done with `relabel_configs` using actions like `labeldrop`, `labelkeep`, `replace`, or `drop`.

Answer

To encrypt metric traffic.

Answer

To increase scrape frequency.

Answer

To store logs in Prometheus.

Question 13

You have two alerts: 'InstanceDown' and 'HighErrorRate'. You want to inhibit 'HighErrorRate' if 'InstanceDown' is firing for the same instance. Where do you configure this logic?

Accepted Answer

In the Alertmanager 'inhibit_rules' configuration. Inhibition rules are defined within the 'inhibit_rules' section of the Alertmanager configuration file.

Answer

In the Prometheus recording rules file

Answer

In the Grafana Alerting panel

Answer

In the Prometheus 'alerting' rule block

Question 14

An engineer executes topk(3, http_requests_total). What happens if there are fewer than 3 time series matching the selector?

Accepted Answer

It returns all available time series without padding or error.. If an instant vector has fewer elements than $k$, topk returns all available elements in the vector without error.

Answer

It pads the remaining slots with synthetic time series valued at 0.

Answer

It returns NaN for the missing slots.

Answer

It returns an evaluation error indicating insufficient time series.

Question 15

Which TWO of the following are valid aggregation operators in PromQL?

Accepted Answer

sum. In PromQL, aggregation operators combine the elements of a single instant vector into a new vector with fewer elements, and the valid set includes sum, min, max, avg, group, stddev, stdvar, count, count_values, bottomk, topk, and quantile. Option A (sum) is correct because sum is a core aggregation operator that adds the values of all series in the vector, optionally grouped by labels via by/without. Option B (stddev) is correct because stddev is a standard aggregation operator that computes the standard deviation across the grouped series. Option C (multiply) is not an aggregation operator; arithmetic binary operators like * (multiply) operate between two instant vectors/scalars rather than aggregating one vector. Option D (filter) is not a PromQL operator at all; filtering is done through label matchers or functions like absent/unless-style logic, not a 'filter' keyword. Option E (rate) is a function, not an aggregation operator, used to calculate the per-second average rate of increase of a counter over a range vector.

Answer

multiply

Answer

filter

Answer

rate