PCA Prometheus Fundamentals Practice Test 1 — 15 Questions

Question 1

You observe data gaps in your Prometheus graph after increasing the scrape interval. What is the most likely cause?

Accepted Answer

The scrape_timeout is shorter than the time taken to fetch metrics. If the scrape interval is too long, the 'scrape_duration' may exceed the interval or cause stale data handling issues.

Answer

TSDB block corruption

Answer

Stale markers not being processed

Answer

Too many labels on the metric

Question 2

You want to rename a metric during the scrape process. Which configuration block is appropriate?

Accepted Answer

metric_relabel_configs. The metric_relabel_configs block in a scrape_config is applied after the scrape and after target relabeling, and it operates on the samples (metrics) themselves — making it the correct place to rename a metric via the 'metric' label or by rewriting __name__. It is specifically designed for post-scrape metric manipulation such as dropping, renaming, or adding labels to metrics.

Answer

scrape_configs

Answer

remote_write

Answer

target_relabel_configs

Question 3

You have a metric that is a gauge and fluctuates. What happens if you use the 'rate()' function on a gauge?

Accepted Answer

It throws an error during query evaluation. rate() is specifically designed for counters; it will not behave as expected on gauges and usually returns an error or empty result.

Answer

It calculates the average change

Answer

It works normally

Answer

It calculates the derivative

Question 4

What label is automatically added by Prometheus to all scraped metrics?

Accepted Answer

instance. Prometheus automatically attaches the 'instance' label to identify the target source.

Answer

region

Answer

job

Answer

host

Question 5

How can you view the current configuration of a running Prometheus server?

Accepted Answer

Visit /config in the UI. The /config endpoint displays the currently loaded configuration.

Answer

Check the /metrics endpoint

Answer

View /status/config

Answer

Run a PromQL query

Answer

Look at the PID file

Question 6

How does Prometheus handle metrics with the same name and identical label sets?

Accepted Answer

It returns a scrape error. Prometheus considers a series unique by its name and label set; duplicate series are not allowed in the same scrape.

Answer

It keeps the first one received

Answer

It averages them

Answer

It keeps both values

Question 7

If you have a scrape_interval of 15s, what is the recommended minimum scrape_timeout?

Accepted Answer

10s. The timeout should generally be slightly less than or equal to the scrape_interval, but it must be sufficient to allow the target to respond.

Answer

5s

Answer

30s

Answer

15s

Question 8

When using 'remote_write', how does Prometheus ensure data delivery?

Accepted Answer

It uses a retry mechanism with exponential backoff. Prometheus uses an internal queue and retry logic to ensure delivery to remote endpoints.

Answer

It writes to the WAL first

Answer

It disables local storage

Answer

It requires manual intervention on failure

Question 9

Where can you see a list of all currently active targets?

Accepted Answer

/targets. The /targets page in the Prometheus UI shows the status of all configured targets.

Answer

/alerts

Answer

/status/targets

Answer

/graph

Question 10

What is the difference between a Histogram and a Summary?

Accepted Answer

Histograms use buckets for quantile estimation. Histograms provide aggregated quantiles on the server side via buckets, while Summaries calculate quantiles on the client side.

Answer

Histograms are computed on the client

Answer

Summaries are computed on the server

Answer

Summaries cannot be aggregated

Question 11

You need to add a static label to all metrics from a specific job. Where do you configure this?

Accepted Answer

scrape_configs.labels. The 'labels' block within a 'scrape_config' adds constant labels to every metric scraped from those targets.

Answer

global.labels

Answer

remote_write.labels

Answer

relabel_configs

Question 12

Which configuration parameter limits the number of series in the head block?

Accepted Answer

storage.tsdb.head-series-limit. The 'storage.tsdb.head-series-limit' flag helps prevent OOM by capping series count.

Answer

max_series

Answer

series_limit

Answer

max_cardinality

Question 13

What is the standard interval for Alertmanager to send notification retries?

Accepted Answer

4h. Alertmanager doesn't have a simple retry interval; it uses the 'repeat_interval' to decide when to re-notify.

Answer

1m

Answer

1s

Answer

24h

Question 14

Which TWO of the following can be modified in 'relabel_configs'?

Accepted Answer

target labels. In Prometheus, relabel_configs operates on the label set of a target before scraping, so it can modify target labels (option B) — for example, using regex replacement to add, drop, or rewrite labels like job or instance. It can also rewrite metric names (option D) because the __name__ label is itself a label, and relabeling rules can set or replace __name__ to rename metrics at ingestion time. Options A, C, and E are not handled by relabel_configs: scrape_interval is set in the scrape_config, retention period is configured via the --storage.tsdb.retention.time flag (or storage.tsdb.retention.size), and alerting logic lives in alerting rules, not in relabeling.

Answer

scrape interval

Answer

retention period

Answer

alerting logic

Question 15

Which THREE factors influence Prometheus memory usage?

Accepted Answer

Number of alerts. Prometheus memory usage is driven primarily by the number of active time series (D), since each series maintains an in-memory head chunk and index entry in the TSDB head block, so cardinality is the dominant factor. The WAL size (C) also matters because the write-ahead log buffers recent samples on disk and its in-memory index/mmap structures scale with the volume of series and samples retained, affecting resident memory. The number of alerts (A) influences memory because each alerting rule evaluation creates and tracks alert state (pending/firing) in memory, and rule evaluations over many series add overhead. The number of dashboards (B) is not a factor because dashboards live in Grafana (or another UI) and only issue queries to Prometheus; they do not consume Prometheus process memory. The scrape interval (E) affects sample ingestion rate and disk/WAL throughput but is not itself a direct driver of Prometheus memory usage the way series cardinality, WAL, and alert state are.

Answer

Number of dashboards

Answer

Scrape interval