> For the complete documentation index, see [llms.txt](https://doc.duaer.com/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://doc.duaer.com/flow-logic/error-handling/memory-errors.md).

# Fix memory issues in Duaer

Duaer does not restrict how much data each node can fetch. When an execution needs more memory than available, it can fail. This page explains how to identify and avoid those errors.
## Identifying out of memory situations

The canvas may show a message such as Execution stopped at this node. When self-hosting, server logs often include Allocation failed - JavaScript heap out of memory.

Typical causes include large JSON item sets, large binary files, many nodes, memory-heavy nodes such as Code, and other workflows running at the same time. Manual executions use more memory because Duaer copies data for the frontend.

## Reduce consumption before you add memory

- Split data into smaller chunks. For example, process 200 rows per execution instead of 10,000. Use [Loop Over Items](/flow-logic/looping.md).
- Avoid the Code node where possible.
- Avoid manual executions when processing larger amounts of data.
- Split into [sub-workflows](/flow-logic/subworkflows.md) and return only a small result set to the parent.

On self-hosted instances you can also raise the process heap, or move executions to workers with [queue mode](/hosting/scaling/queue-mode.md). Refer to [hosting memory](/hosting/scaling/memory-errors.md). Retrying the node that threw still finds a full heap.
## Questions

### What does JavaScript heap out of memory mean in Duaer?

The execution filled the Node.js heap. Common causes are too many items or large binary files in one run. Carry less data first, then raise the heap or use queue mode.

### What should you change first when memory runs out in Duaer?

Reduce data per pass, batch with Loop Over Items, avoid manual runs on large data, and split into sub-workflows when needed. On self-hosted instances, raise the heap or enable queue mode. Retrying the failing node rarely helps.

