Sign in
Home/Splunk/Splunk Enterprise Certified Architect (SPLK-2002)/Free questions

Splunk Enterprise Certified Architect (SPLK-2002) — Free Practice Questions

10 free sample questions from a bank of 161, with the correct answers and explanations. No signup required — start practising right now.

1Which of the following will cause the greatest reduction in disk size requirements for a cluster of N indexers running Splunk Enterprise Security?
  • Setting the cluster search factor to N-1.
  • Increasing the number of buckets per index.
  • Decreasing the data model acceleration range.
  • Setting the cluster replication factor to N-1.
Answer: A

The short version

A — Lowering search factor saves far more disk than lowering replication factor. Searchable copies carry index files plus rawdata while replicas carry only rawdata, so trimming searchables saves most.

Key concepts in this question

  • Replication factor: total number of copies of each bucket the cluster maintains across peer nodes for durability and availability
  • Search factor: number of those copies that are searchable, meaning they include tsidx index files in addition to compressed rawdata
  • Searchable versus non-searchable copies: searchable copies are much larger because tsidx files often approach or exceed rawdata size, especially for Enterprise Security data models
  • Enterprise Security pressure: ES acceleration multiplies tsidx footprint, making search factor dominant

Why A is correct

Setting the cluster search factor to N minus 1 removes one full searchable copy from the retention equation across the cluster. Each searchable copy removed frees far more disk than removing a lean non-searchable copy. Vendor prose confirms replicated non-searchable copies are smaller since they lack associated index files, so reducing search factor has the greater reduction effect. In an ES cluster with heavy index files, this dominates the other options.

Why the others are wrong

  • B. Increasing buckets per index only changes bucket count and roll behavior, not total retained bytes, so disk need stays essentially flat
  • C. Decreasing data model acceleration range helps search performance and acceleration storage, but its savings are smaller than removing a whole searchable replica
  • D. Setting replication factor to N minus 1 removes only a lean non-searchable copy in most layouts, saving mainly compressed rawdata without the large tsidx portion

SPLK-2002 exam tip

Remember search factor costs more disk than replication factor because searchable means tsidx plus rawdata, while replica-only means rawdata alone.

2Which index-time props.conf attributes impact indexing performance? (Select all that apply.)
  • REPORT
  • LINE_BREAKER
  • ANNOTATE_PUNCT
  • SHOULD_LINEMERGE
Answer: B, C

The short version

BC — LINE_BREAKER and ANNOTATE_PUNCT affect indexing performance at index time. They run during parsing on every event, while REPORT runs at search time and SHOULD_LINEMERGE alone is a merge toggle with lighter cost.

Key concepts in this question

  • Index-time versus search-time: index-time props run once per event during ingest and permanently shape storage, so their CPU cost scales with volume
  • LINE_BREAKER: regex-based event boundary detector evaluated against raw streams, expensive when complex or applied to high-volume syslog flows
  • ANNOTATE_PUNCT: punctuation annotation that expands indexed tokens for fast punctuation search, adding write and storage overhead per event
  • REPORT: search-time field extraction alias executed at query time, leaving no index-time parsing burden on indexers
  • SHOULD_LINEMERGE: line-merging switch that controls whether lines combine, far cheaper than full regex breaking or punctuation expansion

Why BC is correct

LINE_BREAKER requires regex evaluation across incoming bytes to decide event splits, which consumes parsing CPU on every kilobyte ingested. ANNOTATE_PUNCT adds extra token metadata per event that must be written into index structures, increasing both CPU and IOPS during indexing. Architecting guidance cited in vendor prose flags these two as performance-relevant index-time attributes. REPORT executes at search time and therefore cannot slow the indexing path, while SHOULD_LINEMERGE is not among the settled performance pair for this question.

Why the others are wrong

  • A. REPORT runs at search time to define field extractions, so it does not burden index-time parsing or bucket writes
  • D. SHOULD_LINEMERGE is a merge behavior toggle, not one of the settled index-time performance drivers selected by the key

SPLK-2002 exam tip

Link index-time cost to LINE_BREAKER plus ANNOTATE_PUNCT, and link REPORT to search time.

3Which of the following are client filters available in serverclass.conf? (Select all that apply.)
  • DNS name.
  • IP address.
  • Splunk server role.
  • Platform (machine type).
Answer: A, B, D

The short version

Banked key AB is incomplete. The correct set is ABD because serverclass.conf supports DNS name, IP address, and platform machine type as client filters, while server role is not a filter.

Key concepts in this question

Server classes group deployment clients for app delivery. Membership is set by filters in forwarder management or serverclass.conf. Splunk defines three kinds: Include, Exclude, and Machine type. Include and Exclude match on IP address, host name, DNS name, or clientName. Machine type matches hardware strings via machineTypesFilter. SPLK-2002 often tests the mapping between UI fields and conf attributes.

Why ABD is correct

A is correct because DNS name is an explicit Include and Exclude value, with wildcards such as star dot ops dot yourcompany dot com. B is correct because IP address is an explicit value, with wildcards such as 10 dot 1 dot 1 dot star. D is correct because platform machine type is a first-class filter via machineTypesFilter, with values such as linux-x86_64 and windows-x64. The spec header also says filters can be based on DNS name, IP address, platform, and clientName.

Why the others are wrong

C is wrong because Splunk server role is never a filtering attribute. Roles describe topology, not client selection. The documented match order is clientName, IP address, DNS name, hostname, and instanceId, plus machineTypesFilter, with no role. Selecting only AB omits a valid filter and fails Select all that apply scoring, so any key without D is incomplete.

SPLK-2002 exam tip

Memorize Include, Exclude, Machine type mapped to whitelist, blacklist, machineTypesFilter. Remember machineTypesFilter needs a whitelist match first, often whitelist dot 0 equals star, and forwarder management always uses filterType whitelist. If server role appears as an option, eliminate it at once.

4What log file would you search to verify if you suspect there is a problem interpreting a regular expression in a monitor stanza?
  • btool.log
  • metrics.log
  • splunkd.log
  • tailing_processor.log
Answer: C

The short version

C — Check splunkd.log for regex interpretation errors in a monitor stanza. It is the primary server log for input pipeline and configuration warnings, while btool only shows merged settings and metrics.log carries performance counters.

Key concepts in this question

  • Monitor stanza inputs: file and directory monitors defined in inputs.conf with whitelist, blacklist, and sourcetype rules that often use regular expressions
  • splunkd.log role: central splunkd operational log recording startup, configuration parsing, input initialization, and WARN or ERROR events from parsing components
  • btool purpose: configuration resolution tool that prints merged effective settings across default and local layers, not runtime errors
  • metrics.log role: periodic throughput, thruput, pipeline, and license usage counters without diagnostics or stack traces

Why C is correct

When a monitor stanza contains a malformed or non-matching regular expression, splunkd logs a parsing or input error at startup or reload time identifying the stanza and pattern problem. Splunk documentation describes splunkd.log as the primary server log requested for troubleshooting, and troubleshooting guides direct regex and monitor issues there. Metrics.log lacks error diagnostics and btool only shows winning values, so splunkd.log directly verifies the problem.

Why the others are wrong

  • A. btool.log does not exist as a runtime diagnostic log and btool output only shows effective config layering without error detail
  • B. metrics.log records periodic performance and license utilization counters, not configuration parse errors or regex warnings
  • D. tailing_processor.log is not the standard consolidated diagnostic log and routine tail status is better confirmed via splunkd.log plus inputstatus endpoints

SPLK-2002 exam tip

Link regex plus monitor to splunkd.log, link counters to metrics.log, and link effective settings to btool.

5Which Splunk tool offers a health check for administrators to evaluate the health of their Splunk deployment?
  • btool
  • DiagGen
  • SPL Clinic
  • Monitoring Console
Answer: D

The short version

D — Monitoring Console is the health-check tool for deployment evaluation. It aggregates topology, performance, license, and search health, while btool, DiagGen, and SPL Clinic are not the admin health dashboard.

Key concepts in this question

  • Monitoring Console: distributed monitoring app that visualizes indexer, search head, forwarder, and license health from a central view
  • Health check scope: resource usage, indexing throughput, search concurrency, disk, and license warnings surfaced as dashboards and alerts
  • btool role: command-line config resolution helper, not a health dashboard
  • Diag artifact: support bundle generator for troubleshooting snapshots, not continuous health evaluation
  • Admin workflow: periodic health review in Monitoring Console guides scaling, rebalancing, and license true-up decisions

Why D is correct

Splunk documentation for the Distributed Management Console describes it as the overview and health interface for administrators to evaluate deployment state. It consolidates search head, indexer, and forwarder status plus license and performance indicators that other tools do not present together. Vendor prose cites the DMC overview and confirms D. Btool only prints effective settings, diag captures a point-in-time bundle, and SPL Clinic is not the health-check answer here. Monitoring Console is therefore the only option matching the health-check description.

Why the others are wrong

  • A. btool resolves merged configuration values but provides no deployment health scoring or trending
  • B. DiagGen style diag creation collects a support snapshot rather than continuous health evaluation
  • C. SPL Clinic is not the documented admin health-check offering for deployment evaluation

SPLK-2002 exam tip

Map health check plus deployment health directly to Monitoring Console.

6In a four site indexer cluster, which configuration stores two searchable copies at the origin site, one searchable copy at site2, and a total of four searchable copies?
  • site_search_factor = origin:2, site1:2, total:4
  • site_search_factor = origin:2, site2:1, total:4
  • site_replication_factor = origin:2, site1:2, total:4
  • site_replication_factor = origin:2, site2:1, total:4
Answer: B

The short version

B — Use site_search_factor origin colon 2, site2 colon 1, total colon 4 for the stated layout. Search factor governs searchable copies, origin and site2 name the constrained sites, and total caps searchable copies cluster-wide.

Key concepts in this question

  • Site search factor: multisite policy controlling how many searchable copies live on each site plus the total searchable count
  • Origin semantics: origin means the site where data originated, allowing two searchable copies to stay local for fast site access
  • Search versus replication: search factor counts searchable tsidx-bearing copies, while replication counts all copies including rawdata-only ones
  • Total constraint: total sets the cluster-wide searchable ceiling so per-site values plus implied remainder sum correctly
  • Four-site context: explicit site names plus total prevent over-replication while satisfying the two plus one plus total-four requirement

Why B is correct

The requirement asks for two searchable copies at origin, one searchable copy at site2, and four searchable copies overall. Only site_search_factor with origin colon 2, site2 colon 1, total colon 4 expresses searchable placement with those exact numbers. Options using site_replication_factor describe total copies rather than searchable copies and miss the searchable guarantee. Option A names the wrong second site. Vendor consensus selects B because search factor is the searchable-copy control the scenario demands.

Why the others are wrong

  • A. Names site1 instead of site2, failing the required placement at site2
  • C. Uses replication factor for a searchable-copy requirement and names the wrong second site
  • D. Uses replication factor, which governs total copies rather than the searchable copies the question specifies

SPLK-2002 exam tip

Read searchable in the stem and immediately choose site_search_factor with matching site names.

7Which Splunk Enterprise offering has its own license?
  • Splunk Cloud Forwarder
  • Splunk Heavy Forwarder
  • Splunk Universal Forwarder
  • Splunk Forwarder Management
Answer: C

The short version

C — Universal Forwarder has its own license model among the listed offerings. It is the lightweight forwarder with distinct licensing.

Key concepts in this question

  • Universal Forwarder: minimal footprint forwarder for reliable data collection with its own license entitlement separate from Enterprise ingest licensing
  • Heavy Forwarder: full Splunk Enterprise instance acting as forwarder with parsing powers, covered under Enterprise licensing rather than a separate forwarder license
  • Cloud forwarder framing: forwarding into Splunk Cloud still follows forwarding license logic, not a distinct own-license product in this option set
  • Forwarder Management: deployment server capability for managing clients, not a separately licensed forwarding product
  • Exam contrast: the question tests which offering stands apart on licensing, making the lightweight forwarder the intended answer

Why C is correct

Splunk licensing distinguishes the Universal Forwarder as a separately licensed lightweight offering that customers deploy widely for collection without consuming Enterprise ingest license in the same way. Heavy forwarders are Enterprise instances performing parsing and routing under Enterprise terms. Forwarder Management is a server role, not a licensed forwarder product. Community and vendor consensus select C as the final answer. Choosing Universal Forwarder therefore matches the only offering with its own license in this set.

Why the others are wrong

  • A. Splunk Cloud Forwarder framing does not represent the separately licensed offering in this option set
  • B. Heavy Forwarder is a full Enterprise instance and follows Enterprise licensing, not its own forwarder license
  • D. Forwarder Management is a management capability, not a separately licensed forwarding product

SPLK-2002 exam tip

Isolate Universal Forwarder as the own-license forwarder among Enterprise-based roles.

8Which component in the splunkd.log will log information related to bad event breaking?
  • Audittrail
  • EventBreaking
  • IndexingPipeline
  • AggregatorMiningProcessor
Answer: D

The short version

D — AggregatorMiningProcessor logs bad event breaking detail in splunkd.log. When line breaking or truncation limits split events incorrectly, this processor component emits the diagnostic message to inspect.

Key concepts in this question

  • Event breaking: parsing-phase splitting of raw streams into discrete events using LINE_BREAKER, SHOULD_LINEMERGE, BREAK_ONLY_BEFORE, and truncation controls
  • Bad event breaking symptoms: merged events, split events, truncated lines, and limit-exceeded warnings such as the 256-character or max-eventsCES style messages
  • splunkd.log component field: each event carries a component name that identifies the emitting processor, enabling filtered troubleshooting searches
  • AggregatorMiningProcessor: parsing component responsible for aggregating and mining event boundaries, cited in troubleshooting guides for breaking errors

Why D is correct

Splunk troubleshooting references and community answers for breaking event because limit has been exceeded point to AggregatorMiningProcessor as the component string to search in splunkd.log. That component handles event boundary detection during aggregation, so misconfigured line breaking surfaces under its name. Filtering by this component isolates the relevant WARN lines, as the vendor reference maps this question to that guide page.

Why the others are wrong

  • A. Audittrail records user actions and audit events, not parsing or line-breaking diagnostics
  • B. EventBreaking looks plausible but is not the actual component string emitted in splunkd.log for this condition
  • C. IndexingPipeline is too generic and points at indexing queue behavior rather than the specific aggregation and breaking processor

SPLK-2002 exam tip

Memorize AggregatorMiningProcessor as the event-breaking signature to grep in splunkd.log.

9Which Splunk server role regulates the functioning of indexer cluster?
  • Indexer
  • Deployer
  • Master Node
  • Monitoring Console
Answer: C

The short version

C — The manager node, formerly called master node, regulates indexer cluster behavior. It coordinates replication, search factor compliance, bucket assignment, and peer recovery from a central control plane.

Key concepts in this question

  • Indexer cluster roles: peer nodes store replicated buckets, manager node orchestrates them, and search heads run distributed searches across peers
  • Manager node duties: tracks bucket copies, assigns primaries, enforces replication and search factors, triggers fixups, and manages rolling restarts
  • Terminology change: master node was renamed to manager node in modern Splunk versions, but exam options may still use the older master label
  • Deployer role: pushes apps and configuration to search head cluster members, with no authority over indexer clustering

Why C is correct

Only the master slash manager node maintains cluster-wide state and instructs peers where to replicate searchable and non-searchable copies. Documentation for enabling the manager node and indexer cluster architecture assigns regulation of the cluster to this role. Indexers themselves are managed peers that follow manager instructions, while deployer and Monitoring Console serve adjacent but distinct functions. This matches legacy master wording and current manager architecture.

Why the others are wrong

  • A. Indexer peers store and serve data but do not regulate replication policy or cluster membership on their own
  • B. Deployer manages search head cluster app distribution, not indexer cluster replication or bucket fixup
  • D. Monitoring Console provides health checks and alerts but has no control-plane authority over clustering

SPLK-2002 exam tip

Map manager equals master equals indexer cluster controller, and deployer equals search head cluster app pusher.

10When adding or rejoining a member to a search head cluster, the following error is displayed: Error pulling configurations from the search head cluster captain; consider performing a destructive configuration resync on this search head cluster member. What corrective action should be taken?
  • Restart the search head.
  • Run the splunk apply shcluster-bundle command from the deployer.
  • Run the clean raft command on all members of the search head cluster.
  • Run the splunk resync shcluster-replicated-config command on this member.
Answer: D

The short version

D — Run splunk resync shcluster-replicated-config on the affected member. The error signals divergent replicated configs, and member-side resync pulls a clean copy from the captain without cluster-wide disruption.

Key concepts in this question

  • Search head cluster replication: captain replicates knowledge objects and configurations to members, keeping search artifacts consistent
  • Pull failure meaning: error pulling configurations indicates the joining member cannot converge with captain state through normal sync
  • Destructive resync: member-side command that wipes local replicated configs and pulls fresh state from the captain for that member only
  • Deployer boundary: deployer pushes apps via shcluster-bundle, which does not repair member-to-captain replicated-config divergence
  • Raft caution: cluster-wide raft cleaning is heavy-handed and inappropriate for a single member sync failure

Why D is correct

Splunk community and docs for this exact error direct admins to perform a destructive configuration resync on the affected member using the resync replicated-config command. That operation clears the member-local replicated state and re-pulls captain content, resolving divergence that restarts alone cannot fix. Vendor prose links the error string to this remedy and selects D. Deployer bundle apply targets deployer-managed apps rather than captain-replicated runtime state, and raft cleaning across all members is excessive. Member-side resync is the precise corrective action.

Why the others are wrong

  • A. Restarting the search head does not replace divergent replicated configs with captain state
  • B. Deployer bundle apply distributes deployer apps, not captain-replicated member configs
  • C. Cleaning raft on all members is cluster-wide and disruptive, not the targeted member resync fix

SPLK-2002 exam tip

Tie pulling-configurations errors to member-side resync shcluster-replicated-config.

Want the full bank of 161 questions for Splunk Enterprise Certified Architect (SPLK-2002)? See all practice exams.