我的问题是:Elasticsearch计数与我的数据库不一样。对嵌套资源使用AND的过滤器
我收录的 “用户” 表中,每个用户可以有一个或多个apps_events:
curl localhost:9200/users/_count
{"count":190291,"_shards":{"total":5,"successful":5,"failed":0}}
SELECT COUNT(*) FROM users
count : 190291
=>相同的计数,一切都很好!
但是,当我做2个过滤器,一个词,一个方面一个嵌套的资源搜索:
curl -X GET 'http://localhost:9200/users/user/_search?load=&size=10&pretty' -d '
{
"query": {
"match_all": {
}
},
"filter": {
"and": [
{
"terms": {
"apps_events.type": [
"sale"
]
}
},
{
"term": {
"apps_events.status": "active"
}
}
]
},
"size": 10
}
total : 63756
而且在我的数据库:
SELECT
COUNT(DISTINCT(users_id))
FROM
apps_event
WHERE
apps_event_state_id = 1 AND apps_event_project_id = 2;
count : 63340
因为实际上,elasticsearch SQL等价查询是:
SELECT
COUNT(DISTINCT(users_id))
FROM apps_event
WHERE apps_event_state_id = 1
AND users_id IN
(SELECT DISTINCT(users_id) FROM apps_event WHERE apps_event_project_id = 2)
count : 63756
===>如何为每个资源做一个简单的“AND”?
感谢
是没有重复的,但是我终于发现,apps_events是一个嵌套的ressource,当你有这样的elasticsearch搜索,在真正的问题: SELECT COUNT (DISTINCT(users_id))FROM apps_event WHERE apps_event_state_id = 1 AND users_id IN(SELECT DISTINCT(users_id)FROM apps_event WHERE apps_event_project_id = 2); 合计:63756 – zywx 2013-02-18 16:30:54
感谢您的跟进。我甚至没有想过嵌套! – drewr 2013-02-18 19:05:04