Elasticsearch 7.10.0
GET repair/_search
{
"size": 0,
"query": {
"bool": {
"must": [
{
"term": {
"typeName": {
"value": "书面",
"boost": 1
}
}
},
{
"range": {
"processStatus": {
"from": null,
"to": 12,
"include_lower": true,
"include_upper": false,
"boost": 1
}
}
},
{
"terms": {
"projectId": [
"499"
],
"boost": 1
}
}
],
"adjust_pure_negative": true,
"boost": 1
}
},
"aggregations": {
"countOrder": {
"terms": {
"field": "projectId",
"size": 732,
"min_doc_count": 1,
"shard_min_doc_count": 0,
"show_term_doc_count_error": false,
"order": [
{
"_count": "desc"
},
{
"_key": "asc"
}
]
},
"aggregations": {
"houseClose": {
"scripted_metric": {
"init_script": {
"source": "state.numas=new HashMap();",
"lang": "painless"
},
"map_script": {
"source": """
if(doc.houseId.size()>0){
String houseKey = doc.houseId.value;
state.numas.put(houseKey,1);
}
""",
"lang": "painless"
},
"combine_script": {
"source": """
double item_finish_count=0;
for(key in state.numas.keySet()){
item_finish_count+=1;
}
return item_finish_count;""",
"lang": "painless"
},
"reduce_script": {
"source": """double result=0;
for(e in states){
result+=e;
}
return result;""",
"lang": "painless"
},
"params": {
"close_sum_key": "close_sum3",
"house_sum_key": "house_sum3"
}
}
},
"houseCount": {
"cardinality": {
"field": "houseId"
}
}
}
}
}
}
结果集：
{
"took" : 392,
"timed_out" : false,
"_shards" : {
"total" : 5,
"successful" : 5,
"skipped" : 0,
"failed" : 0
},
"hits" : {
"total" : {
"value" : 10000,
"relation" : "gte"
},
"max_score" : null,
"hits" : [ ]
},
"aggregations" : {
"countOrder" : {
"doc_count_error_upper_bound" : 0,
"sum_other_doc_count" : 0,
"buckets" : [
{
"key" : "499",
"doc_count" : 17573,
"houseCount" : {
"value" : 1256
},
"houseClose" : {
"value" : 4102.0
}
}
]
}
}
}问题：houseClose和houseCount数据应该一致，但是结果差异很大
这个问题是因为Es针对每个索引是默认建五个分片，所以需要在reduce_script中对结果集进行合并处理去重处理后结果集就对了，为什么要去对Map的key重新计算和去重，场景大概是这样：同一个房间号的key存在不同的分片，然后每个分片的数据源不一样，所以要对这个数据源汇总，再对key去重。