unable to execute the pipeline Novo_ERP_Export via 'Trigger now' or a schedule trigger within synapse spark pool

Question

unable to execute the pipeline Novo_ERP_Export via 'Trigger now' or a schedule trigger within synapse spark pool

Abdulsamad 0

When executing the pipeline Novo_ERP_Export via 'Trigger now' or a schedule trigger within Power Apps, the RunErpExport (SynapseNotebook) activity fails within approximately 4–15 seconds. No Spark application is created, and attempting to open the notebook snapshot in Monitor returns: 'Failed to find the Spark application name from the run output of the activity ('RunErpExport').' This issue is occurring specifically within a Power Apps-related pipeline

Himaja Y 260 Reputation points Microsoft External Staff Moderator

2026-08-18T16:39:30.3333333+00:00

Hi Abdulsamad,

I hope you had a chance to review the information shared earlier, and I hope this information has been helpful! If you still have questions, please let us know what is needed in the comments so the question can be answered.

2 answers

Your answer

Himaja Y 260 Reputation points Microsoft External Staff Moderator

2026-08-18T16:39:30.3333333+00:00

Hi Abdulsamad,

I hope you had a chance to review the information shared earlier, and I hope this information has been helpful! If you still have questions, please let us know what is needed in the comments so the question can be answered.

Answer 1

Hi Abdulsamad,

Based on the symptoms, the failure seems to be happening before the Spark application is created. The “Failed to find the Spark application name” message is likely a secondary error.

Please check the following first:

Go to Synapse/ADF → Monitor → Pipeline runs.
Open the failed Novo_ERP_Export run.
Select RunErpExport → Output and check the result field for the actual error.
If logs are available, check the Driver logs for any ERROR or Exception.

If the output shows a resource or capacity issue, please check the Spark pool/workspace capacity and concurrent job executions. Also verify that the Synapse linked service is configured correctly.

If the pipeline was working previously and no changes were made, it is also worth checking Azure Service Health for any ongoing Synapse Spark issues.

If the issue continues, please share the RunErpExport Output/result error. This will help identify the exact cause.

Answer 2

Use the pipeline run details first, because for pipeline-triggered Synapse Spark failures the inline error is surfaced in the pipeline activity view.

In Synapse/ADF monitoring, open Monitor → Pipeline runs and open the failed run for Novo_ERP_Export.
Select the RunErpExport activity and open Output.
- If the failure is a user error, the result field contains the detailed traceback.
- This is the primary place to inspect when the Spark application was never created.
If the run detail page is available, check the Properties panel for the high-level error and failure stage.
If logs exist, open Logs → Driver logs and search for ERROR, Exception, or out-of-memory.
- If no logs are available, that usually means the job was queued or cluster creation failed, which points to an infrastructure or capacity issue rather than notebook code.

Based on the symptoms described — failure in 4–15 seconds, no Spark application created, and monitor unable to find the Spark application name — the most relevant documented causes to check are:

Workspace or pool capacity/resource shortage
- Error code 3250: not enough resources in the workspace.
- Error code 3251: not enough resources in the pool.
- Error code 3252: not enough vCores for the Spark job.
- Recommended actions: end running jobs, reduce requested vCores, increase workspace quota or pool maximum size, or use another workspace/pool.
Throttling/concurrency pressure
- Error code 3253 indicates throttling due to substantial concurrent executions. The recommendation is to retry after a wait period.
Linked service configuration issue
- Error code 3254 indicates an invalid or missing property in the AzureSynapseArtifacts linked service. Check that required linked service properties are defined correctly.
Notebook structure limitation
- If the notebook used by the pipeline has more than 256 code cells, pipeline execution can fail with error code 6002 and MessageQueueFullException. Reduce the notebook below 256 cells by merging cells.

If this started suddenly and no changes were made, there is also precedent for temporary platform-side Synapse Spark session issues where sessions fail before normal execution. In that case, monitor Azure service health and retry after the incident is mitigated.