pymilvus.exceptions.MilvusException: <MilvusException: (code=2200, message=Retry run out of 75 retry times, message=incomplete query result, missing

Question

What is the following problem?:

pymilvus.exceptions.MilvusException: <MilvusException: (code=2200, message=Retry run out of 75 retry times, message=incomplete query result, missing id %!s(int64=242807), len(searchIDs) = 5, len(queryIDs) = 2, collection=449580455553423281: inconsistent requery result)>

I am trying to do a hybrid search on milvus db. I have two vector embedding fields. based on those to I am trying to do a search using rerank of course. but after plenty time of waiting I am getting the previous error. Any suggestion or any other comment would be great.

UPDATE1: Unfortunately, I can not share code.

• But I have two vector field both of them is sentence transformer embedding, they are dense.

• apart from this two fields, I have other varchar fields.

• I am doing hybrid search on these two vector fields.

• When I set my limit value 1 it is working, no problem. however If I increase the limit number, it is throwing the error that I just shared.

• the important part of the error is the part len(searchIDs) = 5, len(queryIDs) = 2 in which my limit is 5 and it is somehow getting 2 from somewhere I do not know the details. I have tracked your source code, but no chance.

Answer 1

Did you check the doc pages?

• Example multi-vector search when the vectors are all dense.
• Example multi-vector search when mix of sparse and dense vectors.

I've also created an example bootcamp notebook that uses Milvus built-in BGE-M3 embedder to create sparse and dense embeddings with Reranker.

Since you did not include any sample code or logs, we would need to see those to figure out what's going on.

You might want to join our Milvus discord and/or file a github issue and attach your code and error logs to it, to find out more.