WeKnora icon indicating copy to clipboard operation
WeKnora copied to clipboard

[Bug]: 使用多模态模型进行文本分析时模型无响应

Open Zpoteiti opened this issue 1 month ago • 1 comments

相关组件

模型服务

Bug 描述

模型: vllm部署的qwen3 vl 30b a3b instruct 在知识库配置界面测试vlm ocr 能力, 返回测试成功. 但模型后台日志并没有收到任何来自weknora服务器的请求, 测试结果也为空.

Image

期望行为

正确调用模型后返回结果, 若无法ping通模型则返回正确的错误信息.

欢迎与我联系 [email protected]

相关日志

WeKnora-app        | INFO [2025-11-07 15:39:30.360] []                      | [66e6b6b3-62cf-4e15-ae26-37cc907b3857] 200 |  15 |      47.198µs |             ::1 | GET /health
WeKnora-app        | INFO [2025-11-07 15:39:31.607] [request_id=9470ba1b-8858-44a5-bce7-50b2464c1993] tenant.go:84[GetTenantByID] | Start retrieving tenant
WeKnora-app        | INFO [2025-11-07 15:39:31.607] [request_id=9470ba1b-8858-44a5-bce7-50b2464c1993] tenant.go:91[GetTenantByID] | Retrieving tenant, ID: 10000
WeKnora-app        | INFO [2025-11-07 15:39:31.607] [request_id=9470ba1b-8858-44a5-bce7-50b2464c1993] tenant.go:101[GetTenantByID] | Tenant retrieved successfully, ID: 10000, name: admin's Workspace
WeKnora-app        | INFO [2025-11-07 15:39:31.607] [request_id=9470ba1b-8858-44a5-bce7-50b2464c1993] initialization.go:1333[TestMultimodalFunction] | Testing multimodal functionality
WeKnora-app        | INFO [2025-11-07 15:39:31.608] [request_id=9470ba1b-8858-44a5-bce7-50b2464c1993] initialization.go:1366[TestMultimodalFunction] | MinIO config: Bucket=测试, PathPrefix=
WeKnora-app        | INFO [2025-11-07 15:39:31.608] [request_id=9470ba1b-8858-44a5-bce7-50b2464c1993] initialization.go:1378[TestMultimodalFunction] | VLM config: Model=Qwen3-VL-30B-A3B-Instruct-FP8, URL=http://10.180.116.5:6409/v1, HasKey=false, Type=openai
WeKnora-app        | INFO [2025-11-07 15:39:31.608] [request_id=9470ba1b-8858-44a5-bce7-50b2464c1993] initialization.go:1403[TestMultimodalFunction] | Processing image: 屏幕截图 2025-11-07 153116.png, size: 64171 bytes
WeKnora-docreader  | 2025-11-07 07:39:31.608 [9470ba1b-8858-44a5] INFO  __main__             | Received ReadFromFile request for file: 屏幕截图 2025-11-07 153116.png, type: png (elapsed: 0ms)
WeKnora-docreader  | 2025-11-07 07:39:31.609 [9470ba1b-8858-44a5] INFO  __main__             | File content size: 64171 bytes (elapsed: 0ms)
WeKnora-docreader  | 2025-11-07 07:39:31.609 [9470ba1b-8858-44a5] INFO  __main__             | Using chunking config: size=2048, overlap=400, multimodal=True (elapsed: 0ms)
WeKnora-docreader  | 2025-11-07 07:39:31.609 [9470ba1b-8858-44a5] INFO  __main__             | Using Storage config: provider=minio, bucket=测试 (elapsed: 0ms)
WeKnora-docreader  | 2025-11-07 07:39:31.609 [9470ba1b-8858-44a5] INFO  __main__             | Using VLM config: model=Qwen3-VL-30B-A3B-Instruct-FP8, base_url=http://10.180.116.5:6409/v1, interface_type=openai (elapsed: 0ms)
WeKnora-docreader  | 2025-11-07 07:39:31.609 [9470ba1b-8858-44a5] INFO  __main__             | Starting file parsing process (elapsed: 0ms)
WeKnora-docreader  | 2025-11-07 07:39:31.609 [9470ba1b-8858-44a5] INFO  parser.base_parser   | Initializing ImageParser for file: 屏幕截图 2025-11-07 153116.png, type: png (elapsed: 0ms)
WeKnora-docreader  | 2025-11-07 07:39:31.609 [9470ba1b-8858-44a5] INFO  parser.base_parser   | Parser config: chunk_size=2048, overlap=400, multimodal=True, ocr_backend=paddle, max_chunks=1000 (elapsed: 0ms)
WeKnora-docreader  | 2025-11-07 07:39:31.609 [9470ba1b-8858-44a5] INFO  parser.base_parser   | Parsing document with ImageParser, content size: 64171 bytes (elapsed: 0ms)
WeKnora-docreader  | 2025-11-07 07:39:31.609 [9470ba1b-8858-44a5] INFO  parser.image_parser  | Parsing image content, size: 64171 bytes (elapsed: 0ms)
WeKnora-docreader  | 2025-11-07 07:39:31.609 [9470ba1b-8858-44a5] INFO  parser.image_parser  | Uploading image to storage (elapsed: 0ms)
WeKnora-docreader  | 2025-11-07 07:39:31.609 [9470ba1b-8858-44a5] INFO  parser.base_parser   | Uploading bytes content, size: 64171 bytes (elapsed: 0ms)
WeKnora-docreader  | 2025-11-07 07:39:31.609 [9470ba1b-8858-44a5] INFO  parser.storage       | Creating minio storage instance (elapsed: 0ms)
WeKnora-docreader  | 2025-11-07 07:39:31.609 [9470ba1b-8858-44a5] ERROR parser.storage       | Failed to initialize MinIO client: invalid bucket name 测试 (elapsed: 0ms)
WeKnora-docreader  | 2025-11-07 07:39:31.609 [9470ba1b-8858-44a5] INFO  parser.base_parser   | Initialized storage client: MinioStorage (elapsed: 0ms)
WeKnora-docreader  | 2025-11-07 07:39:31.609 [9470ba1b-8858-44a5] INFO  parser.storage       | Uploading bytes content to MinIO, size: 64171 bytes (elapsed: 0ms)
WeKnora-docreader  | 2025-11-07 07:39:31.609 [9470ba1b-8858-44a5] ERROR parser.image_parser  | Failed to upload image to storage (elapsed: 0ms)
WeKnora-docreader  | 2025-11-07 07:39:31.609 [9470ba1b-8858-44a5] INFO  parser.base_parser   | Extracted 0 characters of text from 屏幕截图 2025-11-07 153116.png (elapsed: 0ms)
WeKnora-docreader  | 2025-11-07 07:39:31.609 [9470ba1b-8858-44a5] INFO  parser.base_parser   | Beginning chunking process for text (elapsed: 0ms)
WeKnora-docreader  | 2025-11-07 07:39:31.609 [9470ba1b-8858-44a5] WARNING parser.base_parser   | Empty text provided for chunking, returning empty list (elapsed: 0ms)
WeKnora-docreader  | 2025-11-07 07:39:31.609 [9470ba1b-8858-44a5] INFO  parser.base_parser   | Created 0 chunks from document (elapsed: 0ms)
WeKnora-docreader  | 2025-11-07 07:39:31.609 [9470ba1b-8858-44a5] INFO  parser.base_parser   | Processing images in each chunk for file type: .png (elapsed: 0ms)
WeKnora-docreader  | 2025-11-07 07:39:31.609 [9470ba1b-8858-44a5] INFO  parser.base_parser   | Starting concurrent processing of images in all 0 chunks (elapsed: 0ms)
WeKnora-docreader  | 2025-11-07 07:39:31.609 [9470ba1b-8858-44a5] WARNING parser.base_parser   | No chunks to process (elapsed: 0ms)
WeKnora-docreader  | 2025-11-07 07:39:31.609 [9470ba1b-8858-44a5] WARNING parser.parser        | Parser returned empty chunks for file: 屏幕截图 2025-11-07 153116.png (elapsed: 0ms)
WeKnora-docreader  | 2025-11-07 07:39:31.609 [9470ba1b-8858-44a5] INFO  __main__             | Successfully parsed file 屏幕截图 2025-11-07 153116.png, returning 0 chunks (elapsed: 0ms)
WeKnora-docreader  | 2025-11-07 07:39:31.609 [9470ba1b-8858-44a5] INFO  __main__             | Response size: 0 bytes (elapsed: 0ms)
WeKnora-app        | INFO [2025-11-07 15:39:31.609] [request_id=9470ba1b-8858-44a5-bce7-50b2464c1993] initialization.go:1459[TestMultimodalFunction] | Multimodal test completed successfully in 1ms
WeKnora-app        | INFO [2025-11-07 15:39:31.609] []                      | [9470ba1b-8858-44a5-bce7-50b2464c1993] 200 |  82 |    3.007727ms |   10.180.12.215 | POST /api/v1/initialization/multimodal/test
WeKnora-app        | 2025/11/07 15:39:32.158590 handler.go:31: traces export: rpc error: code = Internal desc = grpc: error while marshaling: string field contains invalid UTF-8

操作系统

Ubuntu 22.04.5 LTS

确认事项

  • [x] 我已经搜索了现有的 issues,确认这是一个新问题

Zpoteiti avatar Nov 07 '25 07:11 Zpoteiti

-- WeKnora-docreader | 2025-11-07 07:39:31.609 [9470ba1b-8858-44a5] ERROR parser.storage | Failed to initialize MinIO client: invalid bucket name 测试 (elapsed: 0ms)

WeKnora-docreader | 2025-11-07 07:39:31.609 [9470ba1b-8858-44a5] ERROR parser.image_parser | Failed to upload image to storage (elapsed: 0ms)

看日志minio写入不了,bucket name 换成英文试试

lyingbug avatar Nov 26 '25 02:11 lyingbug