[Bug]: Knowledge base not referred to when using local Ollama

[teohhc]

Is there an existing issue for the same bug?

• I have checked the existing issues.

Branch name

main

Commit ID

673a28e

Other environment information

vendor_id	: GenuineIntel
cpu family	: 6
model		: 191
model name	: 13th Gen Intel(R) Core(TM) i5-13400F
16GB RAM

Graphics
--------
NVIDIA-SMI 550.67                 
Driver Version: 550.67         
CUDA Version: 12.4
8GB RAM

OS
--
Arch Linux 6.8.5-arch1-1
x86_64

Actual behavior

Ran 2 ollama, one for the embedding (port 11435) and the other for chat (port 11434).

My environment setup

Testing Steps:

1. Ollama models configured without problems
   

2. Change the system model settings
   

3. Knowledge base created
   

4. Retrieval testing without problems
   

5. Chat configuration
   
   
   

6. Chat fails to retrieve from knowledge base
   

Checked the ragflow-logs/api/*.log but can't find any useful hints

Expected behavior

It should refer to the knowledge base for answers similar to the retrieval testing.

Steps to reproduce

I've added phpmyadmin just to check the database.
1. docker-compose up -d ragflow-mysql ragflow-redis ragflow-phpmyadmin ragflow-es-01 minio
2. docker-compose up -d ragflow
3. Refer to https://hub.docker.com/r/ollama/ollama to run both ollama on different ports, llama3 on 11434 and nomic-embed-text (11435)
4. Follow "Testing Steps" give in the "Actual behavior" section

Additional information

Elastic search is still alive and healthy. There is remaining 1GB of free memory left. Both Ollamas didn't report any errors. Retried by clearing docker volumes and recreating the containers and making sure all other services are running first before starting ragflow main service.

[fangxingSR]

The same experience

[DillionApple]

+1

[XiaoCC]

xinference +1

[guoyuhao2330]

同一错误是否存在现有问题？

• 我已经检查了现有问题。

分行名称

主要

提交 ID

673A28E

其他环境信息

vendor_id	: GenuineIntel
cpu family	: 6
model		: 191
model name	: 13th Gen Intel(R) Core(TM) i5-13400F
16GB RAM

Graphics
--------
NVIDIA-SMI 550.67                 
Driver Version: 550.67         
CUDA Version: 12.4
8GB RAM

OS
--
Arch Linux 6.8.5-arch1-1
x86_64

实际行为

运行了 2 个 ollama，一个用于嵌入（端口 11435），另一个用于聊天（端口 11434）。

我的环境设置

测试步骤：

1. Ollama 模型配置没有问题
   
2. Change the system model settings
   
3. Knowledge base created
   
4. Retrieval testing without problems
   
5. Chat configuration
   
   
   
6. Chat fails to retrieve from knowledge base
   

Checked the ragflow-logs/api/*.log but can't find any useful hints

Expected behavior

It should refer to the knowledge base for answers similar to the retrieval testing.

Steps to reproduce

I've added phpmyadmin just to check the database.
1. docker-compose up -d ragflow-mysql ragflow-redis ragflow-phpmyadmin ragflow-es-01 minio
2. docker-compose up -d ragflow
3. Refer to https://hub.docker.com/r/ollama/ollama to run both ollama on different ports, llama3 on 11434 and nomic-embed-text (11435)
4. Follow "Testing Steps" give in the "Actual behavior" section

Additional information

Elastic search is still alive and healthy. There is remaining 1GB of free memory left. Both Ollamas didn't report any errors. Retried by clearing docker volumes and recreating the containers and making sure all other services are running first before starting ragflow main service.

Bug is fixed，please upgrade the dev version.

[teohhc]

I've tested and it is now working as expect.

           _
          |   |
         |   |____
--------/     \____\
       |         \____|
       |        \____|
--------\___\___/