"Private deployment long conversation interruption" #12564

Closed
opened 2026-02-21 19:08:00 -05:00 by yindo · 3 comments
Owner

Originally created by @ASNDTR on GitHub (Apr 5, 2025).

Self Checks

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

Dify version

0.15.3&1.1.3

Cloud or Self Hosted

Self Hosted (Docker)

Steps to reproduce

When privately deploying DeepSeek or Ollama, there is an issue of long conversations being interrupted. When the knowledge base is too large, the prompts are too long, or there are too many turns in the conversation, the response may be interrupted. After the interruption, if the next question has shorter prompts or a smaller knowledge base, and the question is simple, the conversation can continue normally. Additionally, there are no errors in the docker compose logs -f

✔️ Expected Behavior

When using a knowledge base or prompts that are longer (without exceeding the prompt length limit), the conversation proceeds normally without interruption.

Actual Behavior

When the knowledge base is large or contains too much information, the conversation will be interrupted.

Originally created by @ASNDTR on GitHub (Apr 5, 2025). ### Self Checks - [x] This is only for bug report, if you would like to ask a question, please head to [Discussions](https://github.com/langgenius/dify/discussions/categories/general). - [x] I have searched for existing issues [search for existing issues](https://github.com/langgenius/dify/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](https://github.com/langgenius/dify/issues/1542)). - [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&1.1.3 ### Cloud or Self Hosted Self Hosted (Docker) ### Steps to reproduce When privately deploying DeepSeek or Ollama, there is an issue of long conversations being interrupted. When the knowledge base is too large, the prompts are too long, or there are too many turns in the conversation, the response may be interrupted. After the interruption, if the next question has shorter prompts or a smaller knowledge base, and the question is simple, the conversation can continue normally. Additionally, there are no errors in the docker compose logs -f ### ✔️ Expected Behavior When using a knowledge base or prompts that are longer (without exceeding the prompt length limit), the conversation proceeds normally without interruption. ### ❌ Actual Behavior When the knowledge base is large or contains too much information, the conversation will be interrupted.
yindo added the 🐞 bug label 2026-02-21 19:08:00 -05:00
yindo closed this issue 2026-02-21 19:08:00 -05:00
Author
Owner

@dosubot[bot] commented on GitHub (Apr 5, 2025):

Hello @ASNDTR! I'm Dosu, a bot here to help you solve bugs, answer questions, and assist you in becoming a contributor. While you wait for a human maintainer, feel free to share any details or ask anything!

I found some similar discussions that might be relevant to your issue:

  1. Regarding the issue of question and answer interruption - This discussion is still open and might provide insights into similar problems with conversation interruptions [1].

  2. Knowledge-related responses stop in the middle of answers - Another open discussion that deals with interruptions in responses, which could be related to your issue [2].

Additionally, there was a closed issue with a suggested solution that might help:

  • API Connection Issues: If you're running Dify and Ollama inside Docker, ensure 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.

  • Service Status Monitoring: Check the service status at status.deepseek.com to see if there are any ongoing issues with the DeepSeek service.

  • Load Balancing and Error Handling: Implement load balancing or use an error handling node within your workflow to manage errors effectively. The error handling node can provide default values or execute a fail branch to handle exceptions gracefully [3].

These discussions and solutions might provide some guidance on how to address the issue you're experiencing with long conversation interruptions.

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

@dosubot[bot] commented on GitHub (Apr 5, 2025): <!-- Greeting --> Hello @ASNDTR! I'm [Dosu](https://dosu.dev), a bot here to help you solve bugs, answer questions, and assist you in becoming a contributor. While you wait for a human maintainer, feel free to share any details or ask anything! <!-- Answer --> I found some similar discussions that might be relevant to your issue: 1. [Regarding the issue of question and answer interruption](https://github.com/langgenius/dify/discussions/14490) - This discussion is still open and might provide insights into similar problems with conversation interruptions <sup>[[1]](https://github.com/langgenius/dify/discussions/14490)</sup>. 2. [Knowledge-related responses stop in the middle of answers](https://github.com/langgenius/dify/discussions/9379) - Another open discussion that deals with interruptions in responses, which could be related to your issue <sup>[[2]](https://github.com/langgenius/dify/discussions/9379)</sup>. Additionally, there was a closed issue with a suggested solution that might help: - **API Connection Issues**: If you're running Dify and Ollama inside Docker, ensure 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`. - **Service Status Monitoring**: Check the service status at status.deepseek.com to see if there are any ongoing issues with the DeepSeek service. - **Load Balancing and Error Handling**: Implement load balancing or use an error handling node within your workflow to manage errors effectively. The error handling node can provide default values or execute a fail branch to handle exceptions gracefully <sup>[[3]](https://github.com/langgenius/dify/issues/16054)</sup>. These discussions and solutions might provide some guidance on how to address the issue you're experiencing with long conversation interruptions. <!-- Dosu Comment Footer --> *To continue talking to [Dosu](https://dosu.dev), mention @dosu.* --- <sup>Help Dosu learn! Give it feedback: [Great Response](https://app.dosu.dev/response-feedback/7651f039-a186-4257-af6f-bfad39e1fa92?feedback_type=great_response) | [Irrelevant Answer](https://app.dosu.dev/response-feedback/7651f039-a186-4257-af6f-bfad39e1fa92?feedback_type=irrelevant_answer) | [Incorrect Sources](https://app.dosu.dev/response-feedback/7651f039-a186-4257-af6f-bfad39e1fa92?feedback_type=incorrect_sources) | [Too Verbose](https://app.dosu.dev/response-feedback/7651f039-a186-4257-af6f-bfad39e1fa92?feedback_type=too_verbose) | [Hallucination](https://app.dosu.dev/response-feedback/7651f039-a186-4257-af6f-bfad39e1fa92?feedback_type=hallucination) | [Bug Report](https://app.dosu.dev/response-feedback/7651f039-a186-4257-af6f-bfad39e1fa92?feedback_type=bug_report) | [Other](https://app.dosu.dev/response-feedback/7651f039-a186-4257-af6f-bfad39e1fa92?feedback_type=other)</sup>
Author
Owner

@crazywoola commented on GitHub (Apr 5, 2025):

When the knowledge base is too large, the prompts are too long, or there are too many turns in the conversation

DeepSeek series has limited context length. So this is an expected behavior.

@crazywoola commented on GitHub (Apr 5, 2025): > When the knowledge base is too large, the prompts are too long, or there are too many turns in the conversation DeepSeek series has limited context length. So this is an expected behavior.
Author
Owner

@ASNDTR commented on GitHub (Apr 6, 2025):

In my agent, my question is in Chinese, but that doesn’t really matter — as shown in the red circle, the conversation was prematurely interrupted before it finished. But OpenWebUI doesn’t have the issue of conversations getting interrupted midway.

Image

@ASNDTR commented on GitHub (Apr 6, 2025): In my agent, my question is in Chinese, but that doesn’t really matter — as shown in the red circle, the conversation was prematurely interrupted before it finished. But OpenWebUI doesn’t have the issue of conversations getting interrupted midway. ![Image](https://github.com/user-attachments/assets/2a5e5ff0-5d9e-4d3d-bc68-c1cf676f9bbc)
Sign in to join this conversation.
1 Participants
Notifications
Due Date
No due date set.
Dependencies

No dependencies set.

Reference: langgenius/dify#12564