SimFG

@yuege613 How did you deploy milvus, docker or helm? You can try to upload the log file.

@yuege613 ok, I will try to analyze it. thanks your log file

@yuege613 How much data does your milvus service contain? Through the log provided, I found that there are 742 collections. What is your milvus running configuration like? I suspect that...

The current number of collections does not exceed the limit. When you start milvus, are the computer's memory and cpu usage at a high level? What is your computer configuration...

我猜可能是因为你给的这些里面有重复的，我猜的。因为其大概的原理是根据你的输入，去向量数据库里面找类似的内容，然后把这些内容告诉chatgpt，chatgpt在基于你给的这些内容来回答问题。

@highvelcty I tried using your script and no errors were found. And the collection was created correctly 

can you give me more information about the milvus and pymilvus version, and completed error message.

@highvelcty I have found the cause of the problem. A temporary solution is to grant this role the relevant permissions of a collection, for example: ``` root_client.grant_privilege(role_name=USERROLE, object_type='Collection', privilege='Insert', object_name='*')...

@yunurs Sorry for the late reply, I will develop it in the next version.

@yunurs i have merged the pr to the main branch. and you can try to compile this branch to try to use this feature.