Monitoring Fabric Items and Configuring Alerts

What it is

Monitoring answers three questions: did it run, did it succeed, and how much did it cost. Fabric gives a central view in the Monitor hub, detailed views inside each item, query views for warehouses and Eventhouses, and the Capacity Metrics app for capacity usage.

Alerts turn monitoring into action. Activator watches events and data and sends an email or Teams message, or starts a Fabric item, when a condition is met.

How it works

The Monitor hub

The Monitor hub (Monitor in the left navigation) lists recent runs of pipelines, Dataflow Gen2, notebooks, Spark job definitions, semantic model refreshes, Copy jobs and other items you have access to. You filter by status, item type, time and submitter, and open a run to see its details.

Monitor data ingestion

Monitor data transformation

The warehouse keeps query history in query insights views:

SELECT TOP 10 start_time, total_elapsed_time_ms, status, command
FROM queryinsights.exec_requests_history
ORDER BY total_elapsed_time_ms DESC;

Other views include queryinsights.long_running_queries and queryinsights.frequently_run_queries. For queries running now, use sys.dm_exec_requests with sys.dm_exec_sessions.

Workspace monitoring adds a monitoring Eventhouse to a workspace that collects detailed logs and metrics from supported items. You query it with KQL or use its prebuilt dashboards.

Monitor semantic model refresh

A semantic model's Refresh history, in its settings, shows each refresh with start time, duration, status and error text. Refreshes also appear in the Monitor hub. In the scheduled refresh settings you can send refresh failure notifications to the owner and other contacts.

For a Direct Lake model, refresh is a quick metadata operation called framing: the model points at the latest Delta table versions. With automatic updates on, this happens as the tables change. A pipeline can also run a refresh after a load with the Semantic model refresh activity.

Capacity monitoring

The Microsoft Fabric Capacity Metrics app shows capacity unit (CU) consumption by item and operation, and whether the capacity is throttling. Use it to find which notebook, refresh or query is consuming most of the capacity.

Configure alerts

An Activator rule has a condition (such as "status becomes Failed" or "temperature above 80 for 5 minutes") and an action: email, Teams message, running a Fabric item such as a pipeline or notebook, or a Power Automate flow.

A worked example

A nightly pipeline sometimes fails, and the team hears about it only when report users complain the next morning.

They add an Activator rule on the pipeline's job events that sends a Teams message when a run fails, and an Outlook activity on the pipeline's failure path that includes @pipeline().RunId. In the Monitor hub they open the failed run, see that the Copy activity failed, and read its error output.

How the exam asks about this

• "See all pipeline and notebook runs across workspaces in one place." The Monitor hub.
• "Find which items consume the most capacity units." The Fabric Capacity Metrics app.
• "Find the slowest queries run against a warehouse last week." queryinsights.exec_requests_history or queryinsights.long_running_queries.