Try each one before opening the answer. Every option is explained, with the Microsoft documentation page that proves it.
Question 1 · choose 1
A Copy activity that loads a warehouse from an Azure SQL source took three hours last night instead of the usual 40 minutes. Before changing anything, you want to know whether the time went to waiting, reading the source or writing to the destination. Where should you look?
AThe pipeline's schedule settings and the failure notification list
BThe Copy activity's run details, with each step's duration and any tips
CThe warehouse's queryinsights.frequently_run_queries view for the load
DThe Capacity Metrics app's storage page for the copy activity's workspace
Show the answer and why
AThe pipeline's schedule settings and the failure notification list
Incorrect
Schedules and notifications don't break down where a run spent its time.
BThe Copy activity's run details, with each step's duration and any tips
Correct
The run details list each step with its duration, and the longest step is the bottleneck.
CThe warehouse's queryinsights.frequently_run_queries view for the load
Incorrect
This view summarizes repeated SQL queries, not copy steps.
DThe Capacity Metrics app's storage page for the copy activity's workspace
Incorrect
Storage metrics show stored data, not how a copy run spent its time.
Selecting a Copy activity in a pipeline run shows its execution details: data read and written, rows, throughput, the settings used and the steps with their durations, sometimes with performance tuning tips. The step with the longest duration is the bottleneck, which the performance guide then helps you address.
On an overloaded F64 capacity, users' interactive queries and report views are now rejected outright, while scheduled background jobs such as pipelines still start. Earlier in the day the same users only saw short delays. Which throttling phase is the capacity in?
AInteractive delay, because the next 10 minutes of capacity are already used up
BBackground rejection, because the future capacity allowed for background jobs is used up
CInteractive rejection, because the next hour of capacity is already used up
DOverage protection, because usage has been above 100% for under 10 minutes so far
Show the answer and why
AInteractive delay, because the next 10 minutes of capacity are already used up
Incorrect
That earlier phase adds 20-second delays; it doesn't reject interactive requests.
BBackground rejection, because the future capacity allowed for background jobs is used up
Incorrect
In that phase background jobs are rejected too, but here they still start.
CInteractive rejection, because the next hour of capacity is already used up
Correct
Once a full hour of future capacity is consumed, new interactive operations are rejected.
DOverage protection, because usage has been above 100% for under 10 minutes so far
Incorrect
Overage protection doesn't throttle anything at all.
Fabric smooths usage and allows 10 minutes of future capacity without throttling. Beyond that, new interactive operations get 20-second delays; when the next hour is fully used, new interactive operations are rejected while background operations can still start; and in the last phase all new requests are rejected, background ones included.
An eventstream writes to a lakehouse destination, and analysts report that recent events show up in the table later and later. You want to confirm that events are piling up at the destination rather than arriving late from the source. Where should you look?
AThe lakehouse's SQL endpoint query history for the destination's events
BData insights on the lakehouse node, at backlogged input events
CThe Capacity Metrics app's storage page
DThe Spark advisor in a notebook that reads the table
Show the answer and why
AThe lakehouse's SQL endpoint query history for the destination's events
Incorrect
Query history shows SQL reads, not events waiting in the eventstream.
BData insights on the lakehouse node, at backlogged input events
Correct
Destination metrics include backlogged input events, which grow when the destination falls behind.
CThe Capacity Metrics app's storage page
Incorrect
Storage metrics don't show events waiting to be written.
DThe Spark advisor in a notebook that reads the table
Incorrect
The Spark advisor doesn't report on eventstream destinations or their backlog.
An eventstream's Data insights tab shows metrics for the node you select. For a lakehouse destination they include backlogged input events, which tell you whether the destination keeps up with the stream.
An import-mode semantic model built on a warehouse has a scheduled refresh. Last night's refresh failed, and in future the data team (not just the owner) must get an email whenever a refresh fails. Which two steps should you take? (Choose TWO.)
AQuery queryinsights.exec_requests_history for the semantic model's failed refresh
BOpen the semantic model's Refresh history to read the error for the failed attempt
CIn the scheduled refresh settings, add the data team to the failure notification contacts
DTurn on OneLake diagnostics for the workspace
EAdd the data team to a pipeline's scheduled failure notifications
Show the answer and why
AQuery queryinsights.exec_requests_history for the semantic model's failed refresh
Incorrect
Query insights records SQL queries; it doesn't hold the model's refresh errors.
BOpen the semantic model's Refresh history to read the error for the failed attempt
Correct
Refresh history shows the status, duration and error message of each attempt.
CIn the scheduled refresh settings, add the data team to the failure notification contacts
Correct
Failure notifications can go to the owner and other contacts you list.
DTurn on OneLake diagnostics for the workspace
Incorrect
OneLake diagnostics logs data access events, not refresh failures.
EAdd the data team to a pipeline's scheduled failure notifications
Incorrect
This model refreshes on its own schedule, not from a pipeline.
For a semantic model, Refresh > Refresh history lists each attempt with its status, start time, duration and error message. In the scheduled refresh settings you can send refresh failure notifications to the model owner and to other contacts, such as a data team group.
A query that someone else started is blocking loads in a Fabric warehouse. A data engineer with the Member role queries sys.dm_exec_requests but can't see it, and the session must be ended now. Who can find and end it, and how?
AThe Member, after granting herself VIEW SERVER STATE
BAny Contributor, through queryinsights.exec_requests_history
CThe query's author only, by cancelling it from the Monitor hub page
DA workspace Admin, who sees all users' requests and can run KILL
Show the answer and why
AThe Member, after granting herself VIEW SERVER STATE
Incorrect
Members see only their own requests and can't run KILL.
BAny Contributor, through queryinsights.exec_requests_history
Incorrect
Query insights doesn't give a Contributor a way to end another user's session.
CThe query's author only, by cancelling it from the Monitor hub page
Incorrect
Waiting for the author doesn't meet the need to end it now.
DA workspace Admin, who sees all users' requests and can run KILL
Correct
Admins see sessions for all users, and only Admins can run KILL.
In a Fabric warehouse, a workspace Admin can query sys.dm_exec_connections, sys.dm_exec_sessions and sys.dm_exec_requests for all users, while Members, Contributors and Viewers see only their own sessions and requests. Only a workspace Admin can run KILL to stop a session.
An Activator rule watches freezer temperatures from an eventstream. Brief spikes when doors open are normal, so the team must be alerted only when a freezer's temperature stays above -10 for at least 10 minutes. How should the rule be set?
AIs greater than -10, with the occurrence Every time the condition is met
BIs greater than -10, with the occurrence When it has been true for 10 times
CIs greater than -10, with When it has been true for 10 minutes
DNo presence of data for 10 minutes
Show the answer and why
AIs greater than -10, with the occurrence Every time the condition is met
Incorrect
This fires on every brief spike when a door opens.
BIs greater than -10, with the occurrence When it has been true for 10 times
Incorrect
Ten matching events can happen within seconds of one door opening.
CIs greater than -10, with When it has been true for 10 minutes
Correct
This occurrence activates only if the condition stays true for the whole duration.
DNo presence of data for 10 minutes
Incorrect
This heartbeat condition fires when events stop arriving, not when it's warm.
Activator rules combine a condition with an occurrence setting. Every time fires each time the condition is true, a count requires it to be true a number of times, and When it has been true for activates only if the condition remains continuously true for the specified duration, which filters out short spikes.
You want to compare how long each pipeline in a workspace ran last night and spot overlaps between them on one timeline. Which view helps?
AThe Gantt view of all pipeline runs
BThe pipeline's Output tab in the editor
CThe pipeline's schedule settings
DThe workspace's Git status column
Show the answer and why
AThe Gantt view of all pipeline runs
Correct
It shows runs as bars whose length is the run duration.
BThe pipeline's Output tab in the editor
Incorrect
The Output tab shows one run's activities.
CThe pipeline's schedule settings
Incorrect
Schedule settings define runs, not compare them.
DThe workspace's Git status column
Incorrect
Git status shows sync state with the repo.
In the legacy Monitoring hub, the Gantt view displays pipeline runs as bars grouped by pipeline name, with bar length showing run duration, which makes delays and overlaps easy to spot.
Which item watches data for conditions, such as a value crossing a threshold, and then automatically sends an email, a Teams message or starts another item?
AA KQL queryset
BFabric Activator
CA variable library
DA deployment pipeline
Show the answer and why
AA KQL queryset
Incorrect
A queryset is where you write and run KQL queries; it isn't the item that watches data and acts on it.
BFabric Activator
Correct
Activator monitors data, detects conditions and takes actions.
CA variable library
Incorrect
A variable library holds configuration values.
DA deployment pipeline
Incorrect
Deployment pipelines promote content between stages.
Fabric Activator is a no-code tool that connects to data such as eventstreams, detects conditions and patterns, and acts automatically, for example by sending email or Teams messages or starting Fabric items.
A lakehouse destination in an eventstream sometimes drops events, and you want to see the warning and error messages it reports. Which eventstream monitoring view shows them?
AData insights
BThe Real-Time hub catalog
CThe eventstream's Git status
DThe Runtime logs tab
Show the answer and why
AData insights
Incorrect
Data insights doesn't list warning and error messages.
BThe Real-Time hub catalog
Incorrect
The catalog lists streams; it doesn't show node errors.
CThe eventstream's Git status
Incorrect
Git status shows sync state, not runtime errors.
DThe Runtime logs tab
Correct
Runtime logs list warnings and errors for sources and destinations.
Eventstreams offer two monitoring views, Data insights and Runtime logs. The Runtime logs tab shows warning, error and information messages for sources and destinations such as a lakehouse.