MetricSign
EN|NLRequest Access
Low severityexecution

Power BI Refresh Error:
PHOTON_UNSUPPORTED_OPERATION

What does this error mean?

The Databricks Photon engine encountered an operation it does not support and either fell back to the standard Spark engine or raised an error, typically for certain SQL functions or Delta operations.

Common causes

  • 1A SQL function or expression not yet implemented in Photon was used in the query
  • 2The query uses features from a newer Delta protocol version not fully accelerated by the current Photon release
  • 3A Python UDF or Pandas UDF is called within a Photon-accelerated query
  • 4The workload type (e.g., structured streaming with specific triggers) is not Photon-compatible

How to fix it

  1. 1Step 1: Check the Spark UI for the query — Photon fallback is logged as a warning, not always a hard error.
  2. 2Step 2: Identify the specific function causing the fallback using the query plan in the Databricks SQL query history.
  3. 3Step 3: If the fallback is causing significant performance regression, rewrite the operation using Photon-supported functions.
  4. 4Step 4: For Python UDFs, consider converting to native SQL or Scala UDFs which Photon can partially accelerate.
  5. 5Step 5: Monitor the Databricks release notes — Photon coverage expands with each runtime release.

Frequently asked questions

Does Photon fallback mean the query will fail?

No — Photon fallback means the operation runs on the standard Spark engine instead. The query still produces correct results, but without Photon acceleration.

How do I check if Photon is active for my SQL warehouse?

In the Databricks SQL warehouse settings, Photon is enabled by default on Pro and Classic warehouses. Serverless warehouses always use Photon.

Other execution errors