java - SearchHits.getTotalHits()在ElasticSearch中没有给出确切的文档数

我想使用SearchHits.getTotalHits()来获取匹配请求的文档总数。但是，我发现SearchHits.getTotalHits()获得的值不等于SearchHits.getHits().length的总和。正常吗

SearchHits searchHits = response.getHits();
int total = (int)searchHits.getTotalHits();
int k = 0;
do {
    for (SearchHit hit : searchHits.getHits()) {
        k++;
    }
    response = esClient.prepareSearchScroll(response.getScrollId())
            .setScroll(scrollKeepAliveTime)
            .get();
} while (response.getHits().getHits().length != 0);
Assert.assertEquals(total, k);

输出

java.lang.AssertionError: expected:<515> but was:<600>

最佳答案

一个问题是您总是迭代searchHits，即第一个查询的结果。

您需要在每个滚动请求之后将searchHits重新分配给response.getHits()，如下所示:

SearchHits searchHits = response.getHits();
int total = (int)searchHits.getTotalHits();
int k = 0;
do {
    for (SearchHit hit : searchHits.getHits()) {
        k++;
    }
    response = esClient.prepareSearchScroll(response.getScrollId())
            .setScroll(scrollKeepAliveTime)
            .get();
    searchHits = response.getHits();               <-- add this line
} while (searchHits.getHits().length != 0);        <-- modify this line
Assert.assertEquals(total, k);