, Question 1
A Zabbix server is configured with an active proxy that becomes unreachable
for 15 minutes. The proxy's hosts have items with a 60-second update interval
and the proxy buffer is set to 1 hour. Which statement correctly describes data
handling when the proxy reconnects, assuming no item value preprocessing
rejects the data?
A. The proxy discards all buffered values older than the proxy's 'Data
sender frequency' and only sends the most recent value per item.
B. The proxy sends all buffered values; the server stores them with their
original timestamps, and triggers may evaluate historical states in
sequence.
C. The server rejects values older than the item's 'History storage period'
and logs a 'value too old' error for each.
D. The proxy sends only the latest value per item because Zabbix
deduplicates buffered values before transmission.
Correct Answer: B - The proxy sends all buffered values; the
server stores them with their original timestamps, and triggers
may evaluate historical states in sequence.
RATIONALE
Active proxies buffer unsent values and, upon reconnection, transmit
them with original timestamps; the server processes them as historical
data, which can trigger past-state evaluations. Options A and D
incorrectly claim deduplication or discarding, and C misstates the
rejection criterion (history storage period does not block ingestion).
Question 2
Which Zabbix preprocessing step order is required to extract a numeric value
from a JSON payload, convert it to a float, and then apply a custom multiplier,
while ensuring the result is stored as numeric?
A. JSONPath -> Custom multiplier -> Change per second
Page 2
, B. JSONPath -> Change per second -> Custom multiplier
C. Custom multiplier -> JSONPath -> Change per second
D. Change per second -> JSONPath -> Custom multiplier
Correct Answer: A - JSONPath -> Custom multiplier -> Change
per second
RATIONALE
Preprocessing steps execute top-down. JSONPath extracts the value,
then the multiplier scales it, and optional change-per-second can
follow. Placing multiplier before JSONPath (C) fails because the input
is still JSON, and change-per-second before extraction (B/D) would
operate on the raw payload incorrectly.
Question 3
In a Zabbix HA cluster with three nodes, the active node fails. Which
mechanism ensures that only one standby node becomes active, and what is the
typical failover time?
A. Nodes use a shared file lock on the database; failover occurs within 1
minute.
B. Nodes use a distributed consensus protocol (Raft) to elect a leader;
failover is immediate.
C. Nodes compete for a database row lock; the first to acquire it becomes
active, with failover typically under 1 minute.
D. A dedicated HA proxy monitors nodes and promotes a standby via
API; failover takes 5 minutes.
Correct Answer: C - Nodes compete for a database row lock; the
first to acquire it becomes active, with failover typically under 1
minute.
Page 3
, RATIONALE
Zabbix HA uses a database-based locking mechanism where nodes
attempt to acquire a lock row; the first succeeds and becomes active.
Failover is typically under a minute. Options A, B, and D describe
mechanisms not used by Zabbix HA (file locks, Raft, external HA
proxy).
Question 4
A low-level discovery rule returns 10,000 entities. The administrator applies a
filter that discards 90% of them. What is the primary performance impact if the
filter is applied after item prototypes are created, versus before?
A. No difference; Zabbix optimizes discovery regardless of filter order.
B. Applying the filter after creation causes unnecessary item creation and
then deletion, increasing database load and processing time.
C. Applying the filter before creation causes more API calls to the
monitored host.
D. The filter order only affects trigger prototypes, not items.
Correct Answer: B - Applying the filter after creation causes
unnecessary item creation and then deletion, increasing database
load and processing time.
RATIONALE
LLD filters should be applied before creating items to avoid creating
then deleting thousands of items, which burdens the database and
discovery process. Filtering before creation reduces load. Options A,
C, and D misstate the impact.
Question 5
Which statement accurately describes the behavior of a Zabbix calculated item
that references another item's historical data using the 'last' function with a time
shift?
Page 4
A Zabbix server is configured with an active proxy that becomes unreachable
for 15 minutes. The proxy's hosts have items with a 60-second update interval
and the proxy buffer is set to 1 hour. Which statement correctly describes data
handling when the proxy reconnects, assuming no item value preprocessing
rejects the data?
A. The proxy discards all buffered values older than the proxy's 'Data
sender frequency' and only sends the most recent value per item.
B. The proxy sends all buffered values; the server stores them with their
original timestamps, and triggers may evaluate historical states in
sequence.
C. The server rejects values older than the item's 'History storage period'
and logs a 'value too old' error for each.
D. The proxy sends only the latest value per item because Zabbix
deduplicates buffered values before transmission.
Correct Answer: B - The proxy sends all buffered values; the
server stores them with their original timestamps, and triggers
may evaluate historical states in sequence.
RATIONALE
Active proxies buffer unsent values and, upon reconnection, transmit
them with original timestamps; the server processes them as historical
data, which can trigger past-state evaluations. Options A and D
incorrectly claim deduplication or discarding, and C misstates the
rejection criterion (history storage period does not block ingestion).
Question 2
Which Zabbix preprocessing step order is required to extract a numeric value
from a JSON payload, convert it to a float, and then apply a custom multiplier,
while ensuring the result is stored as numeric?
A. JSONPath -> Custom multiplier -> Change per second
Page 2
, B. JSONPath -> Change per second -> Custom multiplier
C. Custom multiplier -> JSONPath -> Change per second
D. Change per second -> JSONPath -> Custom multiplier
Correct Answer: A - JSONPath -> Custom multiplier -> Change
per second
RATIONALE
Preprocessing steps execute top-down. JSONPath extracts the value,
then the multiplier scales it, and optional change-per-second can
follow. Placing multiplier before JSONPath (C) fails because the input
is still JSON, and change-per-second before extraction (B/D) would
operate on the raw payload incorrectly.
Question 3
In a Zabbix HA cluster with three nodes, the active node fails. Which
mechanism ensures that only one standby node becomes active, and what is the
typical failover time?
A. Nodes use a shared file lock on the database; failover occurs within 1
minute.
B. Nodes use a distributed consensus protocol (Raft) to elect a leader;
failover is immediate.
C. Nodes compete for a database row lock; the first to acquire it becomes
active, with failover typically under 1 minute.
D. A dedicated HA proxy monitors nodes and promotes a standby via
API; failover takes 5 minutes.
Correct Answer: C - Nodes compete for a database row lock; the
first to acquire it becomes active, with failover typically under 1
minute.
Page 3
, RATIONALE
Zabbix HA uses a database-based locking mechanism where nodes
attempt to acquire a lock row; the first succeeds and becomes active.
Failover is typically under a minute. Options A, B, and D describe
mechanisms not used by Zabbix HA (file locks, Raft, external HA
proxy).
Question 4
A low-level discovery rule returns 10,000 entities. The administrator applies a
filter that discards 90% of them. What is the primary performance impact if the
filter is applied after item prototypes are created, versus before?
A. No difference; Zabbix optimizes discovery regardless of filter order.
B. Applying the filter after creation causes unnecessary item creation and
then deletion, increasing database load and processing time.
C. Applying the filter before creation causes more API calls to the
monitored host.
D. The filter order only affects trigger prototypes, not items.
Correct Answer: B - Applying the filter after creation causes
unnecessary item creation and then deletion, increasing database
load and processing time.
RATIONALE
LLD filters should be applied before creating items to avoid creating
then deleting thousands of items, which burdens the database and
discovery process. Filtering before creation reduces load. Options A,
C, and D misstate the impact.
Question 5
Which statement accurately describes the behavior of a Zabbix calculated item
that references another item's historical data using the 'last' function with a time
shift?
Page 4
