4.1.1. Monitor Hub: The Central Observatory
💡 First Principle: Monitor Hub provides centralized visibility across all Fabric activities—like a control tower for your data platform. Instead of checking each item individually, you see everything in one place, filtered and sorted for quick triage.
Scenario: It's Monday morning, and you need to verify all weekend batch jobs completed successfully. Without Monitor Hub, you'd check each pipeline individually. With it, you filter by "Failed" status and immediately see problems.
Monitor Hub Capabilities
| Capability | Purpose | Use Case |
|---|---|---|
| Centralized View | All activities in one place | Morning health check |
| Filtering | Status, type, time, user | Troubleshooting failures |
| Drill-down | Examine activity details | Root cause analysis |
| Cross-workspace | See activities across workspaces | Enterprise monitoring |
Filtering Activities
- By Status: Succeeded, Failed, In Progress, Cancelled
- By Type: Pipeline, Dataflow, Notebook, Refresh
- By Time: Custom date range
- By Submitted By: User who triggered the activity
⚠️ Exam Trap: Filter by "Failed" first to quickly identify problems. Looking at all logs without filtering wastes time and buries critical issues in noise.
Run statuses: Not started / Queued (waiting for a concurrent run of the same item or for capacity to free up; not yet running), In progress, Succeeded, Failed (the run did not complete successfully because an activity error was not handled), and Cancelled (stopped by a user before it finished).
Drilling into a run: selecting a pipeline run lists every activity with its status, duration and error message, so a run that took 2 hours instead of 10 minutes shows exactly which activity stalled. A Copy activity's output shows rows read and copied, and a notebook or Spark job links to its Spark application detail page (jobs, stages, executors).
History and failure notifications: the Activities page lists the latest 100 activities from the past 30 days; Historical runs on an item shows that item's full 30-day run history. The Schedule failures page (preview) sets up email notifications, in one place, for when a scheduled run of an item fails (manual runs don't notify; semantic models use their own refresh failure notifications, see 4.1.2).
| Monitoring surface | Answers | Not for |
|---|---|---|
| Monitor hub | Did it run, when, for how long, which activity failed and why, across every workspace you can access | CU consumption, throttling |
| Capacity Metrics app | CU consumption by workspace, item and operation; smoothing, throttling, overload (see 4.1.2) | Per-run activity errors |
| Spark application monitoring | Per-job stages, tasks and executors for tuning a slow notebook | Pipeline control flow |
| Activity log (audit) | Who did what: user and admin actions | Run health |
