dify The DeepSeek thinking process is stuck.

Self Checks

[x] This is only for bug report, if you would like to ask a question, please head to Discussions.
[x] I have searched for existing issues search for existing issues, including closed ones.
[x] I confirm that I am using English to submit this report (我已阅读并同意 Language Policy).
[x] [FOR CHINESE USERS] 请务必使用英文提交 Issue，否则会被关闭。谢谢！:）
[x] Please do not modify this template :) and fill in all the required fields.

Dify version

0.15.3

Cloud or Self Hosted

Self Hosted (Docker)

Steps to reproduce

Ask for help： DeepSeek has been stuck in deep thinking and has not been outputting for a long time. The environment used is: Dify + Ollama (DeepSeek: 32B). Checking the output, it was found that the data has ended.

✔️ Expected Behavior

No response

❌ Actual Behavior

DeepSeek has been stuck in deep thinking and has not been outputting for a long time. The environment used is: Dify + Ollama (DeepSeek: 32B). Checking the output, it was found that the data has ended.

Feb 26 '25 07:02 TaylorSwift5201314

@TaylorSwift5201314 Hello! I'm here to help you with any bugs, questions, or contributions you have for Dify. I'll assist you while we wait for a human maintainer.

A similar issue has been discussed in the Dify repository, where DeepSeek was stuck in deep thinking and not outputting results in the Dify + Ollama (DeepSeek: 32B) environment. Here are some potential solutions to consider:

API Connection Issues: If you are running Dify and Ollama inside Docker, ensure that the network configuration allows proper access. Use 'host.docker.internal' instead of 'localhost' in the Ollama host address configuration, such as http://host.docker.internal:11434 ^[1].
Service Status Monitoring: Check the service status at status.deepseek.com to see if there are any ongoing issues with the DeepSeek service ^[2].
Load Balancing and Error Handling: Implement load balancing or use an error handling node within your workflow to manage errors effectively ^[2].

These steps should help address the issue of DeepSeek being stuck in deep thinking.

To continue talking to Dosu, mention @dosu.

^{Help Dosu learn! Give it feedback: Great Response | Irrelevant Answer | Incorrect Sources | Too Verbose | Hallucination | Bug Report | Other}

Feb 26 '25 07:02 dosubot[bot]

It should be fixed in the latest branch. And you will not see the time duration in v0.15.3. So do what's your version number?

Feb 26 '25 07:02 crazywoola

我的agent也是这种问题，也不调用工具，就直接结束了（版本是：

）

Feb 27 '25 01:02 mu-dan

Previously, I only upgraded the API. Now, I have upgraded both the web and API to version v0.15.3, but the lag issue still persists. The deepseek thinking process ends prematurely. The first time it works normally, but when it comes to the second sentence and needs to combine the context, it always gets stuck.

Feb 28 '25 02:02 TaylorSwift5201314

The issue still persists. The second model output prematurely "message_end", and the output stopped. The environment used is: Dify: V0.15.3 Ollama: V0.5.12 DeepSeek: All products in the series have experienced the premature stoppage of the second thought process output.