PCA Practice Test 21 — 15 Questions

Question 1

Which TWO of the following are true about the Blackbox exporter?

Accepted Answer

It supports HTTP, TCP, and DNS. Option A is correct because the Blackbox exporter performs active probes over protocols such as HTTP/HTTPS, TCP, ICMP, and DNS, exposing results like probe_success and probe_duration_seconds. Option C is correct because the Blackbox exporter is driven by a target list defined in its configuration (e.g., under modules and targets in blackbox.yml or via relabeling in Prometheus), since it needs to know which endpoints to probe. Option B is wrong because the Blackbox exporter is not a database exporter; database metrics are typically collected by dedicated exporters such as mysqld_exporter or postgres_exporter. Option D is wrong because the Blackbox exporter does not instrument application code; that is the role of client libraries or agents like the Prometheus client libraries. Option E is wrong because the Blackbox exporter is an active prober that initiates connections, not a passive collector that only scrapes already-exposed metrics.

Answer

It is the only exporter for databases

Answer

It automatically instruments application code

Answer

It is a passive metric collector

Question 2

Which TWO are true about the 'Summary' metric type?

Accepted Answer

They provide client-side quantiles.. Option C is correct because a Summary metric type computes quantiles (e.g., 0.5, 0.9, 0.99) on the client side, meaning the application itself calculates the quantile values before exposing them, unlike Histograms which expose buckets for server-side quantile estimation. Option D is correct because Summaries have a fixed overhead: the number of quantiles and their sliding time window are configured in advance, so the memory and CPU cost per Summary is predetermined and does not grow with the number of observed values or label combinations. Option A is wrong because Summaries support observations of values (they track sum, count, and quantiles), not only counts. Option B is wrong because client-side quantiles from Summaries cannot be meaningfully aggregated across instances — you cannot average or sum quantiles to get a correct global quantile. Option E is wrong because Summaries and Histograms differ fundamentally: Summaries compute quantiles client-side with fixed overhead, while Histograms expose configurable buckets and allow server-side aggregation and quantile calculation.

Answer

They only support counts.

Answer

They are easily aggregatable across instances.

Answer

They are identical to Histograms.

Question 3

Which TWO of these are true about the 'Pull' model?

Accepted Answer

Prometheus initiates the scrape request.. Option C is correct because in the Pull model Prometheus itself initiates the HTTP scrape request to each target's /metrics endpoint on a schedule, rather than waiting for targets to send data. Option D is correct because, since Prometheus controls the scrape cycle, it can evaluate the `up` metric after each scrape and mark a target as down when the scrape fails or times out, enabling alerting on target availability. Option A is not correct because the pull model's load is predictable and driven by Prometheus's configured scrape intervals and target list, not by unpredictable target behavior. Option B is not correct because targets only need to expose an HTTP metrics endpoint; a push-client is required in push-based models, not in the pull model. Option E is not correct because Prometheus can scrape remote targets over the network, not just local metrics.

Answer

The load on the server is unpredictable.

Answer

Targets must run a push-client.

Answer

It is limited to local metrics.

Question 4

Which THREE aggregation operators in PromQL support the 'by' and 'without' clause modifiers? (Choose three)

Accepted Answer

sum. The three correct answers are A (sum), C (avg), and E (count). These are all PromQL aggregation operators, and every aggregation operator in PromQL — including sum, avg, count, min, max, stddev, stdvar, topk, bottomk, quantile, count_values, and group — accepts the by and without clause modifiers to control how the input vector is grouped before aggregation. For example, sum by (job) (http_requests_total) aggregates per job, while sum without (instance) (http_requests_total) aggregates across all labels except instance. Option B (histogram_quantile) is a function, not an aggregation operator, and it does not take by/without clauses; it operates on buckets produced by an aggregation such as sum by (le) (rate(...)). Option D (label_replace) is a label-manipulation function that rewrites label values and likewise does not support by/without grouping modifiers.

Answer

histogram_quantile

Answer

label_replace

Question 5

Where do you define the scrape interval for a specific target?

Accepted Answer

scrape_configs section. The scrape_interval can be defined within the scrape_config for that specific job.

Answer

global section

Answer

remote_write section

Answer

alerting section

Question 6

Which THREE items are standard techniques to reduce metric cardinality?

Accepted Answer

Aggregating raw metrics into broader categories.. Option A is correct because aggregating raw metrics into broader categories (for example, rolling per-endpoint series up to a per-service or per-method level) collapses many distinct label combinations into fewer time series, directly lowering cardinality. Option C is correct because dropping unused labels via relabeling (e.g., Prometheus metric_relabel_configs with action: labeldrop) removes high-variation or unnecessary label dimensions before ingestion, which is a standard cardinality-reduction technique. Option D is correct because histograms with fewer buckets reduce the number of derived series per histogram, since each bucket becomes its own time series with a le label, so cutting buckets directly cuts cardinality. Option B does not belong because increasing the scrape interval only changes sampling frequency, not the number of distinct label sets, so cardinality is unchanged. Option E does not belong because adding unique IDs to every label does the opposite: it creates a unique series per ID and explodes cardinality.

Answer

Increasing the scrape interval.

Answer

Adding unique IDs to every label.

Question 7

Which TSDB component is responsible for ensuring data durability if the server crashes?

Accepted Answer

WAL (Write-Ahead Log). The Write-Ahead Log (WAL) is a crucial component in Prometheus's TSDB that ensures data durability. Before data is written to the in-memory Head block, it is first appended to the WAL on disk. If the server crashes, the WAL can be replayed to recover any data that was not yet persisted to disk.

Answer

Tombstone file

Answer

Head block

Answer

Index file

Question 8

Which TWO of the following are valid ways to modify or add labels to time series in PromQL? (Choose two)

Accepted Answer

label_replace(). label_replace() (A) is a valid PromQL function that creates or modifies a label on each time series by matching a regex against a source label and writing a replacement value into a destination label, so it can both add and modify labels. label_join() (B) is also a valid PromQL function that concatenates values from multiple source labels into a single destination label, effectively adding or overwriting that label on each series. The remaining options are not real PromQL functions: label_update() (C), label_drop() (D), and label_add() (E) do not exist in the PromQL function set, so they cannot be used to modify or add labels.

Answer

label_update()

Answer

label_drop()

Answer

label_add()

Question 9

You want to find the maximum value of a gauge metric across all instances over a 1-hour sliding window. Which function accomplishes this?

Accepted Answer

max_over_time(node_memory_usage_bytes[1h]). The max_over_time() function evaluates the maximum value of all data points in a specified range vector for each time series.

Answer

max(node_memory_usage_bytes[1h])

Answer

topk(1, node_memory_usage_bytes[1h])

Answer

max_with_range(node_memory_usage_bytes, 1h)

Question 10

Which component is responsible for executing queries written in PromQL?

Accepted Answer

Prometheus Server. The Prometheus server itself contains the query engine that processes PromQL expressions.

Answer

Exporter

Answer

Alertmanager

Answer

Grafana

Question 11

An administrator wants to aggregate CPU usage across all nodes, keeping the instance label while summing the values. Which aggregation clause should be used?

Accepted Answer

sum by (instance) (node_cpu_seconds_total). The by clause specifies the labels that should be preserved in the aggregated result vector.

Answer

sum(node_cpu_seconds_total) keeping (instance)

Answer

sum(node_cpu_seconds_total) group by (instance)

Answer

sum without (instance) (node_cpu_seconds_total)

Question 12

When setting up Alertmanager, what happens if you have no route defined for an alert?

Accepted Answer

The alert is sent to the root route. If an alert matches no specific route, it will fall back to the root route of the configuration tree.

Answer

The alert triggers an error

Answer

The alert is sent to all receivers

Answer

The alert is dropped

Question 13

What is the purpose of the 'le' label in a Histogram?

Accepted Answer

Less than or equal to. The 'le' label stands for 'less than or equal to' and defines the upper bound of the histogram bucket.

Answer

Latency estimate

Answer

Logarithmic exponent

Answer

Last entry

Question 14

When writing a PromQL query involving a vector and a scalar, such as 'node_memory_free_bytes / (1024 * 1024)', how does Prometheus handle label matching?

Accepted Answer

The scalar operation is applied to every time series value in the vector independently, preserving all labels.. Operations between an instant vector and a scalar are performed on each data point of the vector. Labels are unaffected.

Answer

The operation is only applied to the first time series returned by the vector selector.

Answer

The scalar is matched against labels of the vector using exact label equality.

Answer

The query fails because scalars cannot be combined with vectors without an explicit 'on' clause.

Question 15

Which THREE factors contribute to a 'high cardinality' problem in a Prometheus environment?

Accepted Answer

Adding high-variance values like email addresses to labels.. Options A, B, and C are correct because high cardinality in Prometheus is driven by labels whose value sets grow without bound, creating a distinct time series for every unique combination. A is right because high-variance values like email addresses produce a new series per user, exploding the index and memory footprint. B is right because a unique request ID makes essentially every request its own series, which is the classic cardinality anti-pattern. C is right because full URL paths with query parameters multiply series by every distinct path and parameter combination, again unbounded growth. D is not correct because a low scrape interval only increases sample ingestion frequency and storage volume, not the number of distinct series. E is not correct because more scrape targets raise overall series count linearly, but cardinality problems come from high-variance label values, not merely from target quantity.

Answer

Using a low scrape interval.

Answer

Using too many scrape targets.