Skip to content

Debugging a notebook run that failed

In Microsoft Fabric, notebook errors typically surface from Spark job execution issues, code/syntax problems, or resource constraints, and can be diagnosed using the cell-level error output, Spark job monitoring, and the Fabric Monitoring hub. Effective troubleshooting relies on reading stack traces, checking Spark application logs, and correlating notebook runs with underlying Spark session and pool metrics.

Must-know

  • Each notebook cell execution error displays an inline error message with a stack trace; expanding it often reveals the root Spark exception (e.g., AnalysisException, OutOfMemoryError, or Py4JJavaError).
  • The Monitoring hub in Fabric lists all notebook/Spark application runs with status (Succeeded, Failed, Cancelled) and allows drill-down into Spark UI details, logs, and driver/executor diagnostics for failed runs.
  • Common causes of notebook failures include schema mismatches, missing/incorrect library installations, session timeouts, insufficient Spark pool resources, and lakehouse/table path or permission errors.
  • The Spark application detail view (accessible from a notebook run or the Monitoring hub) provides access to driver and executor logs, which are essential for diagnosing memory, shuffle, or executor-lost failures not visible in the notebook cell output itself.
  • High-level severity issues like session expiration or pool capacity exhaustion often require adjusting session timeout settings, Spark pool sizing, or node/executor configuration rather than fixing code.
  • Re-running a failed cell or notebook after a transient error (e.g., a temporary lakehouse connectivity or throttling issue) can resolve it, but recurring failures should be investigated via logs before simply retrying.
Check this objectiveFree · always available

A data engineer builds and tests a notebook in a development workspace that reads files using a relative path such as Files/raw/orders.csv, relying on the lakehouse attached to the notebook. After using deployment pipelines to promote the notebook to a test workspace, running the notebook produces an error stating the specified path cannot be found, even though the equivalent lakehouse and files already exist in the test workspace. What is the most likely cause, and how should the engineer resolve it?

Your objective map0 tried · 0 right · 54 untouched

What you have tried across DP-700's objectives, not a readiness score.

Coverage checked against the published exam guide on Aug 11, 2026.

These are independent practice questions, written against this certification's published exam guide. They are not the certification vendor's own questions, and not the real exam.