PCA Practice Test 18 — 15 Questions

Question 1

You notice that recording rules are consuming too much CPU on your Prometheus server. What is the most effective way to reduce the load while keeping data available?

Accepted Answer

Increase the evaluation_interval. Recording rules in Prometheus are evaluated at the interval defined by 'evaluation_interval' in the rule group. Increasing the evaluation_interval means rules run less frequently, which directly reduces CPU load on the Prometheus server while keeping the recorded time series available for querying.

Answer

Decrease the evaluation_interval

Answer

Move recording rules to Alertmanager

Answer

Delete all indices

Question 2

Which THREE conditions cause a scrape to be unsuccessful?

Accepted Answer

Invalid metric format. In Prometheus, a scrape is unsuccessful when the target cannot be reached or its response cannot be parsed. Option B (Invalid metric format) is correct because Prometheus requires the exposition format to be valid; malformed metric lines, bad labels, or non-numeric values cause the scrape to fail with a parse error. Option C (HTTP 500 status code) is correct because Prometheus treats any non-2xx HTTP response from the /metrics endpoint as a failed scrape, recording it as such in up and scrape_duration_seconds. Option E (Network connection timeout) is correct because if the scrape cannot establish a TCP connection or exceeds the configured scrape_timeout, the scrape fails with a context deadline exceeded error. Options A (High memory usage) and D (Disk space low) do not directly cause a scrape to fail; they may affect the Prometheus server or target host generally, but they are not scrape-level failure conditions recognized by the scrape loop.

Answer

High memory usage

Answer

Disk space low

Question 3

You need to calculate the per-second rate of increase of http_requests_total over a 1-minute window. Why is using a 1-minute range window generally discouraged in production?

Accepted Answer

It is too short relative to typical scrape intervals, leading to high noise and sensitivity to jitter.. Prometheus scrapes most targets every 15 seconds. A 1-minute window only contains about 4 data points, making the rate calculation highly sensitive to small scraping jitters.

Answer

It disables counter reset detection algorithms.

Answer

PromQL automatically rejects any range window shorter than 5 minutes.

Answer

It causes Prometheus to run out of memory due to high sample density.

Question 4

Which Grafana feature allows you to see data from multiple Prometheus servers on a single panel?

Accepted Answer

Data source variables. Grafana allows you to define multiple data sources and even mix them within a dashboard, or use a data source proxy.

Answer

Panel injection

Answer

Metric merging

Answer

Prometheus federation

Question 5

You are joining two vectors where the left vector has labels {job="api", instance="host1"} and the right vector has labels {job="api", region="us-east"}. You use the query 'vector_a + ignoring(instance, region) vector_b'. What does the 'ignoring' modifier do?

Accepted Answer

It causes Prometheus to exclude the listed labels ('instance', 'region') from consideration during the equality check between the two vectors.. The `ignoring` modifier in PromQL vector matching tells Prometheus to ignore the specified labels when matching time series between the two vectors. In this case, `ignoring(instance, region)` means that the labels `instance` and `region` are not considered when finding matching pairs, so the left vector's `instance` and right vector's `region` are excluded from the equality check. This allows matching based on the remaining labels, such as `job`.

Answer

It replaces missing instance and region labels with default values.

Answer

It removes 'instance' and 'region' from the final output time series.

Answer

It hides errors caused by missing 'instance' or 'region' labels in the time series.

Question 6

What is the primary difference between an instant vector and a range vector in PromQL?

Accepted Answer

An instant vector contains a single sample per time series, while a range vector contains a set of samples over time.. Instant vectors contain a single sample per time series for the given timestamp, whereas range vectors contain a set of data points over a time window.

Answer

An instant vector includes timestamps from the future, while range vectors are strictly historical.

Answer

An instant vector can only be used with counters, while range vectors can only be used with gauges.

Answer

A range vector evaluates faster because it caches historical calculations.

Question 7

You are troubleshooting a query and notice that label_replace is being used. What is the primary function of label_replace(v, dst_label, replacement, src_label, regex)?

Accepted Answer

It matches a regex against a source label value and writes the result into a destination label.. The label_replace function matches the regex against the value of src_label, and if it matches, sets dst_label to the replacement string.

Answer

It renames the Prometheus server instance in federated environments.

Answer

It drops all time series that do not contain the specified source label.

Answer

It replaces the metric name with a new string.

Question 8

Which THREE features are provided by the Prometheus HTTP API?

Accepted Answer

Retrieving alert rules. The Prometheus HTTP API exposes endpoints that let clients retrieve alert rules via /api/v1/rules, so option A is correct because it returns configured alerting and recording rules. Option B is correct because the /api/v1/targets endpoint lists active scrape targets along with their labels, state, and last scrape/error information. Option D is correct because the /api/v1/query and /api/v1/query_range endpoints execute PromQL queries against time series data. Option C is not part of the Prometheus HTTP API; SMTP settings for Alertmanager are configured in Alertmanager's configuration file, not through Prometheus's API. Option E is also not provided, since scrape intervals are defined in the Prometheus configuration file (scrape_configs) and require a config reload or restart, not an API call.

Answer

Configuring SMTP

Answer

Updating scrape interval

Question 9

You are seeing 'Too many active series' in your Prometheus logs. Which action should be your first priority to address this?

Accepted Answer

Analyze the number of active series per label name.. Identifying which labels are causing the high cardinality is the standard first step to prevent OOM (Out Of Memory) crashes.

Answer

Switch to a different monitoring tool.

Answer

Disable all alerting rules.

Answer

Increase Prometheus memory limits.

Question 10

Which function evaluates the predictable linear extrapolation of a gauge metric over time to predict future values?

Accepted Answer

predict_linear(node_filesystem_free_bytes[1h], 4 * 3600). predict_linear() is the PromQL function designed to perform linear regression over a range vector and extrapolate the result forward by a specified number of seconds. In predict_linear(node_filesystem_free_bytes[1h], 4 * 3600), Prometheus fits a linear trend to the last hour of samples and projects where the gauge will be in 4 hours, which is the classic pattern for disk-full alerts. The second argument is the extrapolation horizon in seconds, not a duration string.

Answer

holt_winters(node_filesystem_free_bytes[1h], 5m, 1m)

Answer

deriv(node_filesystem_free_bytes[1h])

Answer

extrapolate_linear(node_filesystem_free_bytes[1h], 4 * 3600)

Question 11

Which TWO of the following are valid ways to configure target discovery in Prometheus?

Accepted Answer

dns_sd_configs. Option D, dns_sd_configs, is correct because Prometheus natively supports DNS-based service discovery, allowing it to periodically query A, AAAA, or SRV records to discover targets dynamically. Option E, file_sd_configs, is correct because Prometheus can read target lists from JSON or YAML files on disk and automatically reload them when the files change, making it a valid and commonly used discovery mechanism. The other options are not valid: ftp_sd_configs (A), sql_sd_configs (B), and ldap_sd_configs (C) are not service discovery integrations implemented in Prometheus; Prometheus supports mechanisms such as static_configs, dns_sd_configs, file_sd_configs, and cloud/container SD configs like kubernetes_sd_configs, consul_sd_configs, and ec2_sd_configs, but not FTP, SQL, or LDAP discovery.

Answer

ftp_sd_configs

Answer

sql_sd_configs

Answer

ldap_sd_configs

Question 12

When writing a query with arithmetic operators, how does PromQL handle vector matching when both vectors have the exact same label sets?

Accepted Answer

It performs a one-to-one match automatically based on identical label sets.. When both vectors in a PromQL arithmetic operation have identical label sets, PromQL performs a one-to-one match automatically — each series on the left is paired with the series on the right that has exactly the same labels. No 'on' or 'ignoring' modifier is needed because the default matching behavior already handles identical label sets. This is the foundation of operations like dividing a request-rate metric by a total-rate metric that share the same labels.

Answer

It ignores labels entirely and matches strictly based on metric names.

Answer

It throws a matching error unless the 'on' modifier is explicitly provided.

Answer

It performs a cross-product join of all matching series.

Question 13

Which THREE components are typically involved in the standard Prometheus instrumentation and scraping process?

Accepted Answer

Client library. The standard Prometheus instrumentation and scraping process involves three core components: client libraries (option B), which are used to instrument application code and expose custom metrics via an HTTP endpoint in the Prometheus exposition format; exporters (option D), which translate metrics from third-party systems (e.g., node_exporter, mysqld_exporter) into that same exposition format; and the Prometheus server (option E), which periodically scrapes those HTTP endpoints, stores the time-series data in its TSDB, and evaluates alerting/recording rules. Grafana (option A) is only a visualization layer that queries Prometheus via PromQL and is not part of instrumentation or scraping. Alertmanager (option C) handles alert routing, grouping, and notification after Prometheus evaluates rules, so it is not involved in the instrumentation/scraping pipeline itself.

Answer

Grafana

Answer

Alertmanager

Question 14

Which TWO of the following statements regarding native histograms in Prometheus are correct? (Choose TWO)

Accepted Answer

The histogram_quantile() function can evaluate native histograms directly.. Option B is correct because Prometheus's histogram_quantile() function natively supports native histograms, computing quantiles directly from the stored bucket distribution without needing the classic le-label bucket series. Option C is correct because native histograms encode the entire bucket distribution (counts across dynamically determined buckets) inside a single time series sample, rather than exposing one series per bucket as classic histograms do. Options A, D, and E are incorrect: native histograms use automatically configured, dynamic bucket boundaries so users do not pre-define explicit buckets; they work fine with rate-related PromQL functions like rate() and increase(); and they do not simply double memory consumption of counters, since their overhead depends on bucket count and schema resolution.

Answer

Native histograms require users to manually pre-define explicit bucket boundaries in application code.

Answer

Native histograms are incompatible with all rate-related PromQL functions.

Answer

Native histograms double the memory consumption of traditional counters.

Question 15

Which component is best suited for long-term storage?

Accepted Answer

Remote write storage. Prometheus is not designed for indefinite storage; external remote write storage is needed for long-term retention.

Answer

The local TSDB

Answer

Pushgateway

Answer

Alertmanager