55 questions · Prometheus Fundamentals topic · All types, answers revealed
You are deploying Prometheus in a Kubernetes cluster and need to ensure it discovers pods labeled 'app=frontend'. Which configuration block is required?
kubernetes_sd_configs allows Prometheus to interface with the Kubernetes API for dynamic discovery.
Why this answer
kubernetes_sd_configs is the standard mechanism to discover pods dynamically.
Which label is used to group alerts in Alertmanager?
group_by determines the labels used to bundle alerts.
Why this answer
The 'group_by' field in the route block determines how alerts are grouped into notifications.
If you have a scrape_interval of 15s, what is the recommended minimum scrape_timeout?
It is standard practice to set the timeout to a significant fraction of the interval, often 10s for a 15s interval.
Why this answer
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.
Which THREE conditions cause a scrape to be unsuccessful?
Prevents parsing.
Why this answer
Scrapes fail on network timeout, 404/500 errors, or if the target returns invalid text format.
Which THREE features are provided by the Prometheus HTTP API?
Core functionality.
Why this answer
The API allows querying data, checking target status, and getting rule info.
Which TWO of the following are valid ways to configure target discovery in Prometheus?
Uses DNS lookups.
Why this answer
Prometheus supports multiple SD mechanisms, including file-based and cloud-native providers.
Which component is best suited for long-term storage?
Remote write allows offloading data to long-term storage providers.
Why this answer
Prometheus is not designed for indefinite storage; external remote write storage is needed for long-term retention.
Which THREE of the following metrics are automatically added to all scraped targets?
Indicates scrape success.
Why this answer
Prometheus adds specific metadata labels such as up, scrape_duration, and scrape_samples_scraped.
Which mechanism allows Prometheus to discover targets by reading a list from a file?
file_sd_configs watches files for changes in target definitions.
Why this answer
file_sd_configs is designed to read target lists from JSON or YAML files.
What is the purpose of the 'rule_files' configuration?
rule_files contain the alerting and recording rule logic.
Why this answer
rule_files point to recording and alerting rules that Prometheus evaluates.
What is the primary language used to query Prometheus data?
PromQL is the standard query language for Prometheus.
Why this answer
PromQL (Prometheus Query Language) is the functional query language for Prometheus.
What is the default port that Prometheus listens on?
Prometheus defaults to port 9090.
Why this answer
9090 is the well-known default port for the Prometheus server web interface.
Which TWO of the following labels are specifically associated with Histograms?
Metric suffix for sums.
Why this answer
Histograms use 'le' for buckets and 'sum'/'count' for total metrics.
You want to rename a metric during the scrape process. Which configuration block is appropriate?
metric_relabel_configs allows modifying labels of existing metrics.
Why this answer
relabel_configs is the correct way to modify metric labels (including __name__) before they are ingested.
What label is automatically added by Prometheus to all scraped metrics?
The instance label reflects the target address.
Why this answer
Prometheus automatically attaches the 'instance' label to identify the target source.
Which component is responsible for receiving alerts from Prometheus and managing notification delivery?
Alertmanager handles the full lifecycle of alerts.
Why this answer
Alertmanager is the dedicated component for deduplicating, grouping, and routing alerts.
When a Prometheus server is restarted, how does it recover the in-memory data?
The WAL stores data that hasn't been flushed to disk blocks yet.
Why this answer
It replays the Write-Ahead Log (WAL) to reconstruct the in-memory head block.
What is the benefit of using recording rules?
Precomputing results avoids recalculating them on every query.
Why this answer
Recording rules precompute expensive queries and store them as new series, improving performance.
Your Prometheus server is hitting high memory usage due to high cardinality. Which action is most effective at reducing memory pressure?
Dropping labels prevents high-cardinality metrics from being stored in TSDB.
Why this answer
Reducing label cardinality (e.g., removing unique user IDs) is the primary way to reduce TSDB index size and memory usage.
What is the purpose of 'tombstones' in the TSDB?
Tombstones record that certain data should be ignored.
Why this answer
Tombstones track data points that have been deleted, allowing them to be ignored during queries until they are compacted.
Which THREE of the following are common sources of high cardinality?
High variety of unique strings.
Why this answer
High cardinality often comes from user-provided IDs, timestamps in labels, or dynamic URL parameters.
Which TWO of the following can be modified in 'relabel_configs'?
Standard usage.
Why this answer
Relabeling can modify any label, including the address or metric name.
What is the default value for the scrape interval?
The global default scrape interval is 1 minute.
Why this answer
The default scrape interval in Prometheus is 1 minute.
Which TWO of the following are core responsibilities of the Prometheus server?
The server processes recording and alerting rules.
Why this answer
Prometheus primarily scrapes metrics and evaluates rules.
Where do you define the scrape interval for a specific target?
scrape_configs allows per-job interval configuration.
Why this answer
The scrape_interval can be defined within the scrape_config for that specific job.
Which TSDB component is responsible for ensuring data durability if the server crashes?
The WAL ensures that data in memory can be recovered after a crash.
Why this answer
The Write-Ahead Log (WAL) records incoming samples before they are committed to blocks.
What is the difference between a Histogram and a Summary?
Histograms aggregate counts into predefined buckets.
Why this answer
Histograms provide aggregated quantiles on the server side via buckets, while Summaries calculate quantiles on the client side.
Which operator performs a set union in PromQL?
or returns the union.
Why this answer
The 'or' operator returns the union of two vectors.
What happens if a label set for a metric changes during its lifetime?
Prometheus treats different label sets as entirely different series.
Why this answer
Changing a label set creates a brand new time series in Prometheus.
You are performing a 'label_replace' operation. What is the correct syntax for a regex capture group?
Prometheus uses $1 for the first captured group in replacement strings.
Why this answer
Prometheus uses standard RE2 regex syntax, where capture groups are accessed via $1, $2, etc.
Which configuration parameter limits the number of series in the head block?
This flag limits the active series in memory.
Why this answer
The 'storage.tsdb.head-series-limit' flag helps prevent OOM by capping series count.
In the Prometheus data model, what is a 'sample'?
A sample consists of a float64 value and a millisecond-precision timestamp.
Why this answer
A sample is a value-timestamp pair associated with a specific series.
What is the default retention period for Prometheus data?
15 days is the default storage retention.
Why this answer
The default retention for local Prometheus storage is 15 days.
Which TWO of the following are true about the Pushgateway?
Stale metrics persist until deleted.
Why this answer
Pushgateway is for batch jobs and requires manual deletion of metrics.
When using the 'scrape_timeout', what happens if the timeout is reached?
A timeout results in a failed scrape event.
Why this answer
If the timeout is reached, the scrape is aborted, and a 'up{job=...} == 0' metric is recorded.
When using 'remote_write', how does Prometheus ensure data delivery?
Remote write implements retries to handle temporary network issues.
Why this answer
Prometheus uses an internal queue and retry logic to ensure delivery to remote endpoints.
Which function is used to convert a counter rate into a per-second rate over a time range?
rate() provides the per-second increase of a counter.
Why this answer
rate() is the standard function for per-second rates of counters.
Which THREE of the following are required to expose a new metric in an application?
Essential step.
Why this answer
To expose metrics, you need to define a collector, register it with a registry, and serve it via an HTTP endpoint.
A batch job runs for only 30 seconds every hour. How should you expose these metrics to Prometheus?
The Pushgateway allows transient jobs to push metrics to a persistent location.
Why this answer
The Pushgateway is designed specifically for short-lived jobs that cannot be scraped directly.
Which component allows Prometheus to be scaled horizontally for high availability?
Running multiple identical Prometheus servers is the standard way to achieve HA.
Why this answer
Prometheus instances are typically run in parallel to provide HA; there is no single 'scaling' component.
You observe data gaps in your Prometheus graph after increasing the scrape interval. What is the most likely cause?
If the scrape takes longer than the timeout, the scrape fails, resulting in gaps.
Why this answer
If the scrape interval is too long, the 'scrape_duration' may exceed the interval or cause stale data handling issues.
Which TWO of the following are valid data types in Prometheus?
Standard type.
Why this answer
Prometheus supports Gauges and Counters as primary metric types.
Which TWO of the following can be configured globally?
Global default.
Why this answer
Global configuration includes scrape interval and evaluation interval.
When configuring Alertmanager to send notifications to Slack, which block defines the routing tree?
The route block is the root of the alerting decision tree.
Why this answer
The 'route' block defines the top-level tree for incoming alerts.
You have a metric that is a gauge and fluctuates. What happens if you use the 'rate()' function on a gauge?
The rate() function requires a counter and will error on gauges.
Why this answer
rate() is specifically designed for counters; it will not behave as expected on gauges and usually returns an error or empty result.
Which THREE factors influence Prometheus memory usage?
Alert evaluation consumes memory.
Why this answer
Memory is consumed by the number of series, the size of the WAL, and the number of active alerts.
Which file format does Prometheus use for its main configuration?
Prometheus uses YAML for its primary configuration file.
Why this answer
Prometheus configuration is strictly defined in YAML format.
Which metric type is best for reporting the number of requests received by a service?
Counters track cumulative events.
Why this answer
Counters are monotonic; they only go up, which is perfect for request counts.
Which THREE labels are often considered part of the default Prometheus ecosystem metrics?
Standard for histograms.
Why this answer
While 'job' and 'instance' are standard, 'group' is not a standard automatic label.
Which THREE of the following are components of a Prometheus alert state?
Normal state.
Why this answer
Alerts transition through inactive, pending, and firing states.
Which THREE of the following are valid Prometheus TSDB file types?
Stores compressed time series data.
Why this answer
The TSDB architecture includes chunks, index, and WAL files.
What is the purpose of the 'up' metric?
The 'up' metric reflects scrape success status.
Why this answer
The 'up' metric is 1 if the target was successfully scraped and 0 otherwise.
Which TWO of the following are valid Alertmanager configuration blocks?
Defines the alerting tree.
Why this answer
Alertmanager config includes global settings, route definitions, and receivers.
What is the behavior of the 'absent()' function?
absent() checks for the existence of time series.
Why this answer
absent() returns a vector with a single element if the provided expression has no results.
Which TWO of the following are valid ways to run Prometheus?
Common deployment pattern.
Why this answer
Prometheus can run as a standalone binary or inside a container.
Ready to test yourself?
Try a timed practice session using only Prometheus Fundamentals questions.