特征数据插入问题

[liuruifeng-git]

milvus version 2.0.2
java sdk version 2.0.4

代码：
//插入milvus
List<InsertParam.Field> fields = new ArrayList<>();
fields.add(new InsertParam.Field("features", DataType.FloatVector, featureLists));
fields.add(new InsertParam.Field("date", DataType.Int64, dateLongList));
fields.add(new InsertParam.Field("id", DataType.Int64, dateLongList));
InsertParam insertParam = InsertParam.newBuilder()
.withCollectionName("BidSpaceTest")
.withFields(fields)
.build();
milvusServiceClient.insert(insertParam);

插入了一条数据

数据：[[0.11091897, 0.10981425, 0.10946529, 0.1089742, 0.10914157, 0.108064614, 0.10804585, 0.10828179, 0.10734034, 0.10702038, 0.10663149, 0.10704086, 0.10701183, 0.10654758, 0.10685421, 0.10669913, 0.1064992, 0.10627602, 0.10661164, 0.10728494, 0.10685554, 0.105962515, 0.10686829, 0.10801221, 0.1069447, 0.106962144, 0.106281385, 0.10432063, 0.10262434, 0.10014409, 0.096441545, 0.095015466, 0.09188137, 0.08887424, 0.08656455, 0.08511757, 0.08295897, 0.07980334, 0.07760976, 0.07634725, 0.07645264, 0.075664766, 0.074994124, 0.07825248, 0.07656363, 0.07601002, 0.07278369, 0.070964634, 0.07016153, 0.06848421, 0.06809411, 0.06753272, 0.066838585, 0.06698062, 0.06736541, 0.068205856, 0.069764845, 0.07153347, 0.0736938, 0.07581258, 0.074329704, 0.07707215, 0.080937155, 0.08457597, 0.09076262, 0.098599695, 0.10443987, 0.10933307, 0.11352261, 0.116187304, 0.11698627, 0.120516226, 0.1222908, 0.12383664, 0.12682448, 0.12900138, 0.13056545, 0.13194276, 0.1324476, 0.13245995, 0.13170451, 0.12993094, 0.12835029, 0.12715496, 0.1254788, 0.12357853, 0.121472396, 0.12026113, 0.117506854, 0.11533012, 0.112836026, 0.110219866, 0.10879817, 0.10778891, 0.107614845, 0.1063989]]

attu查看库中的值：
[0.10946529358625412,0.10897420346736908,0.10914157330989838,0.10806461423635483,0.10804585367441177,0.10828179121017456,0.10734034329652786,0.10702037811279297,0.10663148760795593,0.10704085975885391,0.1070118322968483,0.10654757916927338,0.10685420781373978,0.1066991314291954,0.10649920254945755,0.1062760204076767,0.10661163926124573,0.10728494077920914,0.10685554146766663,0.10596251487731934,0.10686828941106796,0.10801220685243607,0.10694470256567001,0.10696214437484741,0.10628138482570648,0.10432063043117523,0.10262434184551239,0.10014408826828003,0.09644154459238052,0.09501546621322632,0.09188137203454971,0.08887424319982529,0.0865645483136177,0.08511757105588913,0.08295896649360657,0.07980334013700485,0.07760976254940033,0.0763472467660904,0.07645264267921448,0.07566476613283157,0.07499412447214127,0.07825247943401337,0.07656362652778625,0.07601001858711243,0.0727836936712265,0.07096463441848755,0.07016152888536453,0.06848420947790146,0.06809411197900772,0.06753271818161011,0.0668385848402977,0.06698062270879745,0.06736540794372559,0.06820585578680038,0.06976484507322311,0.07153347134590149,0.07369379699230194,0.07581257820129395,0.07432970404624939,0.0770721510052681,0.08093715459108353,0.08457597345113754,0.09076262265443802,0.09859969466924667,0.1044398695230484,0.10933306813240051,0.11352261155843735,0.11618730425834656,0.11698626726865768,0.12051622569561005,0.12229079753160477,0.12383663654327393,0.1268244832754135,0.12900137901306152,0.13056544959545135,0.1319427639245987,0.13244760036468506,0.13245995342731476,0.13170450925827026,0.12993094325065613,0.12835028767585754,0.12715496122837067,0.1254788041114807,0.12357853353023529,0.12147239595651627,0.12026113271713257,0.11750685423612595,0.11533012241125107,0.11283602565526962,0.11021986603736877,0.10879816859960556,0.10778891295194626,0.10761484503746033,0.10639890283346176,3.3160085836446607e-38,0]

用attu csv导入后的值：
[0.11091896891593933,0.10981424897909164,0.10946529358625412,0.10897420346736908,0.10914157330989838,0.10806461423635483,0.10804585367441177,0.10828179121017456,0.10734034329652786,0.10702037811279297,0.10663148760795593,0.10704085975885391,0.1070118322968483,0.10654757916927338,0.10685420781373978,0.1066991314291954,0.10649920254945755,0.1062760204076767,0.10661163926124573,0.10728494077920914,0.10685554146766663,0.10596251487731934,0.10686828941106796,0.10801220685243607,0.10694470256567001,0.10696214437484741,0.10628138482570648,0.10432063043117523,0.10262434184551239,0.10014408826828003,0.09644154459238052,0.09501546621322632,0.09188137203454971,0.08887424319982529,0.0865645483136177,0.08511757105588913,0.08295896649360657,0.07980334013700485,0.07760976254940033,0.0763472467660904,0.07645264267921448,0.07566476613283157,0.07499412447214127,0.07825247943401337,0.07656362652778625,0.07601001858711243,0.0727836936712265,0.07096463441848755,0.07016152888536453,0.06848420947790146,0.06809411197900772,0.06753271818161011,0.0668385848402977,0.06698062270879745,0.06736540794372559,0.06820585578680038,0.06976484507322311,0.07153347134590149,0.07369379699230194,0.07581257820129395,0.07432970404624939,0.0770721510052681,0.08093715459108353,0.08457597345113754,0.09076262265443802,0.09859969466924667,0.1044398695230484,0.10933306813240051,0.11352261155843735,0.11618730425834656,0.11698626726865768,0.12051622569561005,0.12229079753160477,0.12383663654327393,0.1268244832754135,0.12900137901306152,0.13056544959545135,0.1319427639245987,0.13244760036468506,0.13245995342731476,0.13170450925827026,0.12993094325065613,0.12835028767585754,0.12715496122837067,0.1254788041114807,0.12357853353023529,0.12147239595651627,0.12026113271713257,0.11750685423612595,0.11533012241125107,0.11283602565526962,0.11021986603736877,0.10879816859960556,0.10778891295194626,0.10761484503746033,0.10639890283346176]

[yhmo]

@liuruifeng-git
I could not reproduce this issue.
My test steps:

1. start milvus v2.0.2
2. use java sdk v2.0.4 to create a collection and insert 1 row
   the client code:

void test() {
        String randomCollectionName = "test_1";
        int dimension = 96;
        
        // collection schema
        String field1Name = "id";
        String field2Name = "feature";
        String field3Name = "date";
        List<FieldType> fieldsSchema = new ArrayList<>();
        fieldsSchema.add(FieldType.newBuilder()
                .withPrimaryKey(true)
                .withAutoID(false)
                .withDataType(DataType.Int64)
                .withName(field1Name)
                .build());

        fieldsSchema.add(FieldType.newBuilder()
                .withDataType(DataType.FloatVector)
                .withName(field2Name)
                .withDimension(dimension)
                .build());

        fieldsSchema.add(FieldType.newBuilder()
                .withDataType(DataType.Int64)
                .withName(field3Name)
                .build());

        // create collection
        CreateCollectionParam createParam = CreateCollectionParam.newBuilder()
                .withCollectionName(randomCollectionName)
                .withFieldTypes(fieldsSchema)
                .build();

        R<RpcStatus> createR = client.createCollection(createParam);

        // insert data
        Long[] idArr = {999L};
        List<Long> ids = Arrays.asList(idArr);

        Float[] featureArr = {0.11091897f, 0.10981425f, 0.10946529f, 0.1089742f, 0.10914157f, 0.108064614f, 0.10804585f, 0.10828179f, 0.10734034f, 0.10702038f, 0.10663149f, 0.10704086f, 0.10701183f, 0.10654758f, 0.10685421f, 0.10669913f, 0.1064992f, 0.10627602f, 0.10661164f, 0.10728494f, 0.10685554f, 0.105962515f, 0.10686829f, 0.10801221f, 0.1069447f, 0.106962144f, 0.106281385f, 0.10432063f, 0.10262434f, 0.10014409f, 0.096441545f, 0.095015466f, 0.09188137f, 0.08887424f, 0.08656455f, 0.08511757f, 0.08295897f, 0.07980334f, 0.07760976f, 0.07634725f, 0.07645264f, 0.075664766f, 0.074994124f, 0.07825248f, 0.07656363f, 0.07601002f, 0.07278369f, 0.070964634f, 0.07016153f, 0.06848421f, 0.06809411f, 0.06753272f, 0.066838585f, 0.06698062f, 0.06736541f, 0.068205856f, 0.069764845f, 0.07153347f, 0.0736938f, 0.07581258f, 0.074329704f, 0.07707215f, 0.080937155f, 0.08457597f, 0.09076262f, 0.098599695f, 0.10443987f, 0.10933307f, 0.11352261f, 0.116187304f, 0.11698627f, 0.120516226f, 0.1222908f, 0.12383664f, 0.12682448f, 0.12900138f, 0.13056545f, 0.13194276f, 0.1324476f, 0.13245995f, 0.13170451f, 0.12993094f, 0.12835029f, 0.12715496f, 0.1254788f, 0.12357853f, 0.121472396f, 0.12026113f, 0.117506854f, 0.11533012f, 0.112836026f, 0.110219866f, 0.10879817f, 0.10778891f, 0.107614845f, 0.1063989f};
        List<List<Float>> features = new ArrayList<>();
        features.add(Arrays.asList(featureArr));

        Long[] dateArr = {20200602L};
        List<Long> dates = Arrays.asList(dateArr);

        List<InsertParam.Field> fieldsInsert = new ArrayList<>();
        fieldsInsert.add(new InsertParam.Field(field1Name, DataType.Int64, ids));
        fieldsInsert.add(new InsertParam.Field(field2Name, DataType.FloatVector, features));
        fieldsInsert.add(new InsertParam.Field(field3Name, DataType.Int64, dates));

        InsertParam insertParam = InsertParam.newBuilder()
                .withCollectionName(randomCollectionName)
                .withFields(fieldsInsert)
                .build();

        R<MutationResult> insertR = client.insert(insertParam);

        // get collection statistics
        R<GetCollectionStatisticsResponse> statR = client.getCollectionStatistics(GetCollectionStatisticsParam
                .newBuilder()
                .withCollectionName(randomCollectionName)
                .withFlush(true)
                .build());
        assertEquals(R.Status.Success.getCode(), statR.getStatus().intValue());

        GetCollStatResponseWrapper stat = new GetCollStatResponseWrapper(statR.getData());
        System.out.println("Collection row count: " + stat.getRowCount());
    }

3. start Attu service
   docker run -p 8000:3000 -e HOST_URL=http://{ machine IP }:8000 -e MILVUS_URL={ machine IP}:19530 zilliz/attu:latest

4. load the collection in Attu, query the vector by "id == 999", the result is:
   [0.11091896891593933,0.10981424897909164,0.10946529358625412,0.10897420346736908,0.10914157330989838,0.10806461423635483,0.10804585367441177,0.10828179121017456,0.10734034329652786,0.10702037811279297,0.10663148760795593,0.10704085975885391,0.1070118322968483,0.10654757916927338,0.10685420781373978,0.1066991314291954,0.10649920254945755,0.1062760204076767,0.10661163926124573,0.10728494077920914,0.10685554146766663,0.10596251487731934,0.10686828941106796,0.10801220685243607,0.10694470256567001,0.10696214437484741,0.10628138482570648,0.10432063043117523,0.10262434184551239,0.10014408826828003,0.09644154459238052,0.09501546621322632,0.09188137203454971,0.08887424319982529,0.0865645483136177,0.08511757105588913,0.08295896649360657,0.07980334013700485,0.07760976254940033,0.0763472467660904,0.07645264267921448,0.07566476613283157,0.07499412447214127,0.07825247943401337,0.07656362652778625,0.07601001858711243,0.0727836936712265,0.07096463441848755,0.07016152888536453,0.06848420947790146,0.06809411197900772,0.06753271818161011,0.0668385848402977,0.06698062270879745,0.06736540794372559,0.06820585578680038,0.06976484507322311,0.07153347134590149,0.07369379699230194,0.07581257820129395,0.07432970404624939,0.0770721510052681,0.08093715459108353,0.08457597345113754,0.09076262265443802,0.09859969466924667,0.1044398695230484,0.10933306813240051,0.11352261155843735,0.11618730425834656,0.11698626726865768,0.12051622569561005,0.12229079753160477,0.12383663654327393,0.1268244832754135,0.12900137901306152,0.13056544959545135,0.1319427639245987,0.13244760036468506,0.13245995342731476,0.13170450925827026,0.12993094325065613,0.12835028767585754,0.12715496122837067,0.1254788041114807,0.12357853353023529,0.12147239595651627,0.12026113271713257,0.11750685423612595,0.11533012241125107,0.11283602565526962,0.11021986603736877,0.10879816859960556,0.10778891295194626,0.10761484503746033,0.10639890283346176]

[zhansg]

简单说就是attu建的表有bug，插入的标量、向量字段都会被改变。java sdk建的表没问题