mirror of
https://github.com/langgenius/dify-official-plugins.git
synced 2026-07-21 17:45:23 -04:00
Adjusted 'timeout': Timeout (615.0, read=600.0, write=10.0, connect=5.0), "max_retries": 1, Not effective #296
Reference in New Issue
Block a user
Delete Branch "%!s()"
Deleting a branch is permanent. Although the deleted branch may continue to exist for a short time before it actually gets removed, it CANNOT be undone in most cases. Continue?
Originally created by @kuku-jiusgan on GitHub (May 19, 2025).
Self Checks
Dify version
1.1.3
Plugin version
0.0.16
Cloud or Self Hosted
Self Hosted (Docker)
Steps to reproduce
当模型300S内没有输出时，会这样报错，但当我修改了 credentials_kwargs = {
"api_key": credentials['openai_api_key'],
"timeout": Timeout(615.0, read=600.0, write=10.0, connect=5.0),
"max_retries": 3,
}中的read为600后，没有生效
✔️ Error log
[openai_api_compatible] Error: PluginInvokeError: {"args":{},"error_type":"ConnectionError","message":"HTTPConnectionPool(host='10.0.0.111', port=10002): Read timed out."}
@dosubot[bot] commented on GitHub (Aug 17, 2025):
Hi, @kuku-jiusgan. I'm Dosu (https://dosu.dev), and I'm helping the dify-official-plugins team manage their backlog and am marking this issue as stale.
Issue Summary:
Next Steps:
Thank you for your understanding and contribution!
@1107186916 commented on GitHub (Aug 18, 2025):
Has it been solved? I have the same problem with the Qwen2.5-72B-Instruct-128K model of siliconflow.
my dify is 1.7.1
SiliconFlow plugins is 0.0.15
@dosubot[bot] commented on GitHub (Aug 18, 2025):
@crazywoola, could you please assist the user with their ongoing timeout issue related to the Qwen2.5-72B-Instruct-128K model in their Dify setup? They mentioned using Dify 1.7.1 and SiliconFlow plugin 0.0.15.
@shihuimin1012 commented on GitHub (Sep 12, 2025):
最终问题不解决，就一直关闭issues，你是官方吗？超时这个300，改什么都没用，好多人遇到了，这代表只要用本地模型的很多都会不好用，还不解决？
@kuku-jiusgan commented on GitHub (Sep 12, 2025):
我自己已经解决，这个报错是插件的报错，要修改插件模块中的一个超时设置，但是特别麻烦，在docker环境里
@dosubot[bot] commented on GitHub (Sep 12, 2025):
@crazywoola 用户反馈最终问题未解决，超时300设置无效，且本地模型使用时仍有问题，尤其是在docker环境中，请帮忙跟进处理。
@shihuimin1012 commented on GitHub (Sep 12, 2025):
openai_api_compatible 你用的也是这个吗 请问最终如何解决的啊 我这一直超时
@shihuimin1012 commented on GitHub (Sep 12, 2025):
@kuku-jiusgan
@kuku-jiusgan commented on GitHub (Sep 12, 2025):
比较麻烦，进插件的docker环境，有个llm.py文件，进去找到超时相关的设置，把300改成你想要的时间就行
@shihuimin1012 commented on GitHub (Sep 16, 2025):
请问是进入插件的docker容器内有这个代码目录，修改嘛。我修改了，未起作用；
@shihuimin1012 commented on GitHub (Sep 16, 2025):
@kuku-jiusgan
@kuku-jiusgan commented on GitHub (Sep 16, 2025):
@shihuimin1012 commented on GitHub (Sep 16, 2025):
我用的openai_api_compatible插件，和你最初的修改地方一样。但是没起作用。
我修改的models/openai_api_compatible/models/common_openai.py 文件；
当模型300S内没有输出时，会这样报错，但当我修改了 credentials_kwargs = {
"api_key": credentials['openai_api_key'],
"timeout": Timeout(615.0, read=600.0, write=10.0, connect=5.0),
"max_retries": 3,
}中的read为600后，没有生效
为何你最后修改了llm。py文件 是改用ollma了嘛
@kuku-jiusgan commented on GitHub (Sep 16, 2025):
不是，llm.py是Dify的llm.py文件，需要进到容器里修改
这是之前也是git上有个伙计问我当时的截图，然后他的路径，你可以按照这个路径找一下
@shihuimin1012 commented on GitHub (Sep 16, 2025):
我找找，感谢；
也就是说除了修改，models/openai_api_compatible/models/common_openai.py 这个文件，还需要修改llm.py 才可以搞定这个超时是啊？而不是官方说的llm.py是ollma插件的。
@kuku-jiusgan commented on GitHub (Sep 16, 2025):
@shihuimin1012 是的，但是看文件路径，确实是open_api下的。 我觉得应该是ollama下也有个llmpy文件，所以 机器人有点混淆
@shihuimin1012 commented on GitHub (Sep 16, 2025):
感谢感谢，我修改成功了，终于不是5min就timeout了
@dosubot[bot] commented on GitHub (Oct 2, 2025):
Hi, @kuku-jiusgan. I'm Dosu, and I'm helping the dify-official-plugins team manage their backlog and am marking this issue as stale.
Issue Summary:
Next Steps:
Thank you for your understanding and contribution!