What causes OutOfMemoryException?

Lookup activity retrieving a very large result set without row limits configured. Multiple concurrent pipeline activities running on the same underpowered IR node. IR node machine has insufficient RAM for the data volume being processed. Memory leak in a long-running Self-Hosted IR process that has not been restarted

How do I fix OutOfMemoryException?

Step 1: Check the memory usage on the Self-Hosted IR host machine using Task Manager or Performance Monitor; confirm available RAM is critically low during the failure window.. Step 2: If the error occurs in a Lookup activity, add a row limit or apply a query filter to reduce the result set size returned into memory.. Step 3: Scale up the IR node machine by adding more RAM, or distribute load by registering additional nodes in the same IR pool.. Step 4: Restart the Self-Hosted IR service on the affected node to release any accumulated memory and verify all IR nodes are online before re-running the pipeline.. Step 5: Enable IR diagnostic logs via the Self-Hosted IR Configuration Manager and upload them to Microsoft Support if the issue persists after scaling.

High severitycapacityAzure Data Factory →

Power BI Refresh Error:
OutOfMemoryException

What does this error mean?

The Self-Hosted Integration Runtime node ran out of available memory while executing an activity, most commonly during Lookup or Copy operations that attempt to load large datasets into RAM.

Common causes

1Lookup activity retrieving a very large result set without row limits configured
2Multiple concurrent pipeline activities running on the same underpowered IR node
3IR node machine has insufficient RAM for the data volume being processed
4Memory leak in a long-running Self-Hosted IR process that has not been restarted

How to fix it

1Step 1: Check the memory usage on the Self-Hosted IR host machine using Task Manager or Performance Monitor; confirm available RAM is critically low during the failure window.
2Step 2: If the error occurs in a Lookup activity, add a row limit or apply a query filter to reduce the result set size returned into memory.
3Step 3: Scale up the IR node machine by adding more RAM, or distribute load by registering additional nodes in the same IR pool.
4Step 4: Restart the Self-Hosted IR service on the affected node to release any accumulated memory and verify all IR nodes are online before re-running the pipeline.
5Step 5: Enable IR diagnostic logs via the Self-Hosted IR Configuration Manager and upload them to Microsoft Support if the issue persists after scaling.

Frequently asked questions

Can I limit memory usage per activity in ADF rather than scaling up the machine?

ADF does not expose a per-activity memory cap, but you can reduce memory pressure by paginating large queries, using Data Flow instead of Lookup for bulk data, or moving the workload to an Azure IR where memory is managed elastically.

Will adding more Self-Hosted IR nodes fix this error?

Yes, if the OOM is caused by concurrent activities overwhelming a single node. ADF distributes jobs across online nodes in the same IR, so adding nodes reduces per-node load. However, if a single activity alone exceeds available RAM, you must also reduce dataset size or increase the per-node RAM.