[Bug]: full text seach failed with error Out of range in json: param 'bm25_avgdl' (0.969) should be in range [1.000000, 340282346638528859811704183484516925440.000000]

[zhuwenxing]

Is there an existing issue for this?

• I have searched the existing issues

Environment

- Milvus version:master
- Deployment mode(standalone or cluster):
- MQ type(rocksmq, pulsar or kafka):    
- SDK version(e.g. pymilvus v2.0.0rc2):
- OS(Ubuntu or CentOS): 
- CPU/Memory: 
- GPU: 
- Others:

Current Behavior

[2025/01/24 17:34:55.154 +08:00] [INFO] [base/milvus_client.go:32] [Response] [method=Search] [resp="{\"status\":{\"error_code\":1,\"reason\":\"fail to search on QueryNode 1: worker(1) query failed: N6milvus21ExecOperatorExceptionE :Operator::GetOutput failed for [Operator:, plan node id: 517] :  =\\u003e failed to search: config={{\\\"bm25_avgdl\\\":0.969,\\\"k\\\":5,\\\"metric_type\\\":\\\"BM25\\\",\\\"nprobe\\\":\\\"16\\\",\\\"s......"]
    response_checker.go:21: 
                Error Trace:    /Users/zilliz/workspace/milvus/tests/go_client/common/response_checker.go:21
                                                        /Users/zilliz/workspace/milvus/tests/go_client/testcases/full_text_search_test.go:269
                Error:          Received unexpected error:
                                fail to search on QueryNode 1: worker(1) query failed: N6milvus21ExecOperatorExceptionE :Operator::GetOutput failed for [Operator:, plan node id: 517] :  => failed to search: config={{"bm25_avgdl":0.969,"k":5,"metric_type":"BM25","nprobe":"16","span_id":"36396a9cf1c74d69","trace_flags":0,"trace_id":"55a7d8367324607625eab8491d60ab2f"}} out of range in json: Out of range in json: param 'bm25_avgdl' (0.969) should be in range [1.000000, 340282346638528859811704183484516925440.000000] at /workspace/source/internal/core/src/index/VectorMemIndex.cpp:412
                Test:           TestSearchFullTextWithEmptyData
[2025/01/24 17:34:55.155 +08:00] [INFO] [base/milvus_client.go:32] [Request] [method=DropCollection] [reqs="{\"collection_name\":\"full_text_search_empty\"}"]
[2025/01/24 17:34:55.169 +08:00] [INFO] [base/milvus_client.go:32] [Response] [method=DropCollection] [resp={}]
--- FAIL: TestSearchFullTextWithEmptyData (3.83s)

// TestSearchFullTextWithEmptyData tests full text search with empty data
func TestSearchFullTextWithEmptyData(t *testing.T) {
	ctx := hp.CreateContext(t, time.Second*common.DefaultTimeout)
	mc := createDefaultMilvusClient(ctx, t)

	params := hp.NewFullTextParams().
		TWithCollectionName("full_text_search_empty").
		TWithTextFields([]string{"text"})

	var schema *entity.Schema
	schema, err := hp.DefaultFullTextPrepare.CreateCollection(ctx, t, mc, params)
	common.CheckErr(t, err, true)

	err = hp.DefaultFullTextPrepare.CreateIndex(ctx, t, mc, schema, "0.2")
	common.CheckErr(t, err, true)

	loadTask, err := mc.LoadCollection(ctx, milvusclient.NewLoadCollectionOption(schema.CollectionName))
	common.CheckErr(t, err, true)
	err = loadTask.Await(ctx)
	common.CheckErr(t, err, true)

	// Insert data with high empty percentage
	option := hp.NewFullTextDataOption().
		TWithLanguage("english").
		TWithEmptyPercent(80)

	err = hp.DefaultFullTextPrepare.InsertDataWithOption(ctx, t, mc, schema, 1000, option)
	common.CheckErr(t, err, true)

	flushTask, err := mc.Flush(ctx, milvusclient.NewFlushOption(schema.CollectionName))
	common.CheckErr(t, err, true)
	err = flushTask.Await(ctx)
	common.CheckErr(t, err, true)

	results, err := hp.DefaultFullTextPrepare.Search(ctx, t, mc, schema,
		"test query", 5, []string{"id", "text"})
	common.CheckErr(t, err, true)
	require.Equal(t, 1, len(results))
}

Expected Behavior

No response

Steps To Reproduce

Milvus Log

No response

Anything else?

No response

[zhuwenxing]

when empty percent is 50, this error was not reproduced

[zhuwenxing]

/assign @zhengbuqian

[yanliang567]

/unassign

[zhengbuqian]

@zhuwenxing thanks! this should have been fixed by zilliztech/knowhere#1077, please verify in Milvus branch 2.5, with the latest knowhere with tag v2.5.3.

[zhengbuqian]

/assign @zhuwenxing

[zhuwenxing]

can stable reproduce in master-20250127-b3791a6f-amd64
after fix
not reproduced in master-20250217-a4dbbc2e-amd64 and 2.5-20250219-f904b736-amd64 (both images tag should contains fix)