n8n: Execution stopped at this node (n8n may have run out of memory)
Last verified
Execution stopped at this node (n8n may have run out of memory while executing it)What it means
n8n shows this in the editor when an execution died while a node was running, and the most likely reason is that the n8n process ran out of memory. The docs list it alongside "Problem running workflow", "Connection Lost" and "503 Service Temporarily Unavailable", which suggest the whole instance became unavailable.
n8n keeps every item that passes through a workflow in memory during an execution. Large JSON datasets, large binary files (PDFs, images, video), many nodes, memory-heavy nodes such as Code, and manual runs (which copy data for the editor) all add up. Other executions running at the same time share the same memory.
On n8n Cloud the instance restarts automatically and memory depends on your plan. On self-hosted Docker it restarts automatically; on npm installs you may need to restart it manually.
What it looks like
Main workflow: fetch IDs -> Loop Over Items (batch 200) -> Execute Workflow (process batch) -> loop
Sub-workflow returns only { "processed": 200 }Why it happens
- Processing thousands of items or a very large API response in one execution.
- Large binary files held in memory.
- Code nodes that build big arrays or strings, such as Base64-encoding a file.
- Running a heavy workflow manually from the editor.
- Several heavy executions at once on a small instance.
How to fix it
- Process data in smaller chunks, for example 200 rows per execution instead of 10,000, as n8n's docs suggest.
- Move per-batch work into a sub-workflow (Loop Over Items + Execute Workflow) that returns only a small result.
- Avoid the Code node where a built-in node can do the job.
- Avoid manual runs with large data; test with a small sample.
- Self-hosted: give the instance more RAM; n8n Cloud: upgrade to a plan with more memory.
How to stop it recurring
Design workflows to stream through data in batches and to store binary data outside memory where your setup allows. Watch instance memory and execution sizes. The underlying server log message is JavaScript heap out of memory.
Official documentation: docs.n8n.io/hosting/scaling/memory-errors. See all n8n error messages or the n8n limits and quotas.
Related errors
Definitions
Quick answers
What does "Execution stopped at this node (n8n may have run out of memory)" mean in n8n?
n8n shows this in the editor when an execution died while a node was running, and the most likely reason is that the n8n process ran out of memory. The docs list it alongside "Problem running workflow", "Connection Lost" and "503 Service Temporarily Unavailable", which suggest the whole instance became unavailable. n8n keeps every item that passes through a workflow in memory during an execution.
How do I fix "Execution stopped at this node (n8n may have run out of memory)"?
1. Process data in smaller chunks, for example 200 rows per execution instead of 10,000, as n8n's docs suggest. 2. Move per-batch work into a sub-workflow (Loop Over Items + Execute Workflow) that returns only a small result. 3. Avoid the Code node where a built-in node can do the job. 4. Avoid manual runs with large data; test with a small sample. 5. Self-hosted: give the instance more RAM; n8n Cloud: upgrade to a plan with more memory.
Error handling you can see in the workflow
Autonoly workflows have per-node retries, timeouts, continue-on-fail and error-handler branches, so a failing step can be retried or routed instead of stopping the whole run. Saved workflows have no loop node; lists are worked through by the AI agent or one run per trigger item.
Also see Visual workflow builder.