langflow-ai/langflow · error · HTTPException

Error getting knowledge base.

Error message

Error getting knowledge base.

What it means

Catch-all 500 from the get-single-KB endpoint: exceptions while resolving the KB path, loading metadata from disk (knowledge_base_service.load_metadata_from_disk), or computing directory size are logged with the kb_name and wrapped. HTTPExceptions (e.g. 404 for a missing KB) pass through unchanged.

Source

Thrown at src/backend/base/langflow/api/v1/knowledge_bases.py:1489

                dir_name=record.name,
                metadata=knowledge_base_service.record_to_metadata_dict(record),
                size=record.size_bytes,
            )

        kb_path = _resolve_kb_path(kb_name, _kb_guard.owner_user)
        metadata = knowledge_base_service.load_metadata_from_disk(kb_path)
        return _build_kb_info(
            kb_name=kb_name.replace("_", " "),
            dir_name=kb_name,
            metadata=metadata,
            size=KBStorageHelper.get_directory_size(kb_path),
        )

    except HTTPException:
        raise
    except Exception as e:
        await logger.aerror("Error getting knowledge base '%s': %s", kb_name, e)
        raise HTTPException(status_code=500, detail="Error getting knowledge base.") from e


@router.get("/{kb_name}/chunks", status_code=HTTPStatus.OK, dependencies=[Depends(_check_memory_base_association)])
async def get_knowledge_base_chunks(
    kb_name: str,
    current_user: CurrentActiveUser,
    request: Request,
    page: Annotated[int, Query(ge=1)] = 1,
    limit: Annotated[int, Query(ge=1, le=100)] = 50,
    search: Annotated[str, Query(description="Filter chunks whose text contains this substring")] = "",
    source_type: Annotated[
        str | None,
        Query(description="Only return chunks ingested via the given source type (e.g. 'file_upload', 'folder')."),
    ] = None,
    file_name: Annotated[
        str | None,
        Query(description="Only return chunks whose source filename exactly matches."),
    ] = None,

View on GitHub (pinned to 976ec789d2)

Solutions

1. Check server logs for "Error getting knowledge base '<kb>': <e>" to see whether it was metadata parsing or size computation.
2. Verify the KB directory on the server is intact and readable.
3. If the directory is gone, delete the stale KB record and recreate the KB.
4. Restore the metadata file from backup or re-save the KB config via the UI.

async def kb_gettable(client, kb_name: str) -> bool:
    return (await client.get(f"/api/v1/knowledge_bases/{kb_name}")).status_code == 200

try:
    resp = await client.get(f"/api/v1/knowledge_bases/{kb_name}")
except HTTPStatusError as e:
    if e.response.status_code == 500:
        log("KB metadata/size read failed", kb_name)  # check server log
    elif e.response.status_code == 404:
        log("KB not found", kb_name)

When it happens

Trigger: GET /api/v1/knowledge_bases/{kb_name} where the KB row exists but metadata loading fails (corrupted/unreadable metadata file) or KBStorageHelper.get_directory_size raises (permission errors, broken symlinks, deleted directory still recorded in DB).

Common situations: KB files deleted or moved behind the backend's back, NFS/permission changes on the storage root, or partially written metadata after a crash.

Related errors

• Error ingesting files to knowledge base.
• Error ingesting folder to knowledge base.
• Error listing knowledge bases.
• Error listing metadata keys.
• Metadata value for '{key}' exceeds {KB_METADATA_MAX_VALUE_LE

AI-assisted analysis of langflow-ai/langflow@976ec789d2 (2026-08-14). Data as JSON: /api/errors/434d3d7dcde4d118. Report an issue: GitHub.