Bug Description

Bug: Qdrant Vector Store / Tool Vector Store fails with NodeOperationError: terminated and no actionable error details

Description

I am experiencing a persistent failure when using a Qdrant Vector Store through an n8n AI Tool Vector Store node.

The issue happens during execution of an AI Agent using a Tool Vector Store connected to:

• Qdrant Vector Store node
• Ollama Embeddings node
• Tool Vector Store node
• AI Agent / Tools Agent V3

The workflow fails when the Tool Vector Store tries to query the Qdrant Vector Store.

The only error reported by n8n is:

terminated

There is no additional diagnostic information in the n8n UI or n8n logs beyond terminated.

Expected behavior

If the Qdrant Vector Store query fails, n8n should expose a more actionable error message, for example:

• Qdrant HTTP status code
• Qdrant response body
• connection error
• timeout reason
• aborted request reason
• invalid filter / payload schema error
• embedding-related error
• upstream LangChain error details

At the moment the error is too generic to debug.

Actual behavior

The execution fails with:

NodeOperationError: terminated

The error object contains:

{
  "description": "terminated",
  "message": "terminated",
  "messages": [],
  "context": {},
  "functionality": "configuration-node",
  "name": "NodeOperationError"
}

No Qdrant response body, HTTP status code, request details, or lower-level cause is exposed.

Technical setup

Relevant node types and versions:

@n8n/n8n-nodes-langchain.agent
Type version: 3.1

@n8n/n8n-nodes-langchain.toolVectorStore
Type version: 1.1

@n8n/n8n-nodes-langchain.vectorStoreQdrant
Type version: 1.3

@n8n/n8n-nodes-langchain.embeddingsOllama
Type version: 1

The Tool Vector Store is configured in retrieve mode and connected to a Qdrant Vector Store node.

The Qdrant Vector Store uses a Qdrant collection and the searchFilterJson option.

Example structure:

{
  "qdrantCollection": "collection_name",
  "options": {
    "searchFilterJson": "{...}"
  }
}

The filter is a valid Qdrant-style payload filter using must, for example:

{
  "must": [
    {
      "key": "metadata.user_id",
      "match": {
        "value": "some_user_id"
      }
    }
  ]
}

The embeddings are provided by an Ollama Embeddings node.

n8n log behavior

I checked the n8n logs with debug logging enabled.

The relevant sequence is:

debug | Start executing node "<Tool Vector Store node>"
debug | Running node "<Tool Vector Store node>" started
debug | Supply data for embeddings Ollama
debug | Running node "<Tool Vector Store node>" finished with error
debug | Workflow execution finished with error

The failed node runs for approximately 2 minutes before failing.

The final error in the n8n log is:

NodeOperationError: terminated

There are no additional messages in the error object:

{
  "messages": [],
  "context": {},
  "description": "terminated",
  "message": "terminated"
}

Relevant stack trace

Sanitized stack trace from n8n logs:

NodeOperationError: terminated
    at ExtendedQdrantVectorStore.callMethodAsync (.../@n8n/ai-utilities/src/utils/log-wrapper.ts:53:17)
    at processTicksAndRejections (node:internal/process/task_queues:104:5)
    at Proxy.<anonymous> (.../@n8n/ai-utilities/src/utils/log-wrapper.ts:467:25)
    at VectorDBQAChain._call (.../@langchain/classic/src/chains/vector_db_qa.ts:76:18)
    at VectorDBQAChain.invoke (.../@langchain/classic/src/chains/base.ts:125:24)
    at VectorDBQAChain.run (.../@langchain/classic/src/chains/base.ts:218:26)
    at VectorStoreQATool.call (.../@langchain/core/src/tools/index.ts:310:19)
    at ExecuteContext.execute (.../@n8n/n8n-nodes-langchain/nodes/tools/ToolVectorStore/ToolVectorStore.node.ts:153:23)
    at WorkflowExecute.executeNode (.../n8n-core/src/execution-engine/workflow-execute.ts:1048:9)
    at WorkflowExecute.runNode (.../n8n-core/src/execution-engine/workflow-execute.ts:1239:11)

This suggests the error is thrown while the Tool Vector Store is calling the Qdrant-backed retriever through LangChain.

Additional technical observations

• The error occurs inside the Tool Vector Store / Qdrant Vector Store integration path.
• The AI Agent itself starts correctly.
• The embeddings node is supplied before the failure.
• The failure is surfaced as a NodeOperationError.
• The underlying cause is not exposed.
• The error contains no Qdrant-specific diagnostic data.
• The n8n logs do not contain any additional information beyond terminated.
• The stack trace points to ExtendedQdrantVectorStore.callMethodAsync and then LangChain VectorDBQAChain.

Why this is hard to debug

The current error message does not make it possible to determine whether the failure is caused by:

• Qdrant timeout
• Qdrant connection reset
• invalid filter JSON
• unsupported payload filter
• embedding dimension mismatch
• collection schema/vector mismatch
• request abort
• LangChain timeout
• n8n internal timeout
• upstream HTTP client termination

Since the error only says terminated, there is no clear next step for troubleshooting.

Suggested improvement

Please expose the underlying error cause from the Qdrant / LangChain layer when this happens.

Useful information would include:

- original error name
- original error message
- cause
- Qdrant HTTP status code, if available
- Qdrant response body, if available
- request timeout / abort reason, if applicable
- sanitized Qdrant endpoint path
- whether the error occurred during embedding, retrieval, filtering, or response parsing

Even a wrapped error like this would be much more useful:

QdrantVectorStore query failed: <underlying error>

instead of only:

terminated

Minimal reproduction pattern

A minimal workflow should be enough to reproduce the issue:

1. AI Agent node
2. Tool Vector Store node
3. Qdrant Vector Store node
4. Ollama Embeddings node
5. Qdrant collection with payload metadata
6. searchFilterJson using a must filter on a metadata field
7. Agent invokes the Tool Vector Store

The failure occurs when the Tool Vector Store queries the Qdrant Vector Store.

Environment

n8n deployment: Docker/self-hosted
Qdrant: external/self-hosted Qdrant API
Embeddings: Ollama embeddings
AI Agent: Tools Agent V3
Vector store: Qdrant Vector Store
Tool: Tool Vector Store

Exact n8n version can be provided if needed.

Main issue

The main bug is not only that the Qdrant query fails, but that n8n reports only:

terminated

with no actionable technical details, making the failure effectively opaque.

To Reproduce

The issue happens during execution of an AI Agent using a Tool Vector Store connected to:

• Qdrant Vector Store node
• Ollama Embeddings node
• Tool Vector Store node
• AI Agent / Tools Agent V3

The workflow fails when the Tool Vector Store tries to query the Qdrant Vector Store.

The only error reported by n8n is:

terminated

Expected behavior

Expected behavior

If the Qdrant Vector Store query fails, n8n should expose a more actionable error message, for example:

• Qdrant HTTP status code
• Qdrant response body
• connection error
• timeout reason
• aborted request reason
• invalid filter / payload schema error
• embedding-related error
• upstream LangChain error details

At the moment the error is too generic to debug.

Debug Info

Debug info

core

• n8nVersion: 2.18.4
• platform: docker (self-hosted)
• nodeJsVersion: 24.14.1
• nodeEnv: production
• database: sqlite
• executionMode: regular
• concurrency: -1
• license: enterprise (production)
• consumerId: ...

storage

• success: all
• error: all
• progress: false
• manual: true
• binaryMode: filesystem

pruning

• enabled: true
• maxAge: 336 hours
• maxCount: 10000 executions

client

• userAgent: mozilla/5.0 (windows nt 10.0; win64; x64) applewebkit/537.36 (khtml, like gecko) chrome/147.0.0.0 safari/537.36
• isTouchDevice: false

Generated at: 2026-05-04T19:38:12.901Z

Operating System

Debian LXC

n8n Version

2.18.4

Node.js Version

24.14.1

Database

SQLite (default)

Execution mode

main (default)

Hosting

self hosted