- springboot版本:2.0.5.RELEASE
- elasticsearch版本:7.9.1
1、配置
引入依赖:
<dependency>
<groupId>org.elasticsearch.client</groupId>
<artifactId>elasticsearch-rest-high-level-client</artifactId>
<version>7.9.1</version>
</dependency>
<dependency>
<groupId>org.elasticsearch</groupId>
<artifactId>elasticsearch</artifactId>
<version>7.9.1</version>
</dependency>
application.properties 配置文件:
elasticsearch.schema=http
elasticsearch.address=192.168.80.130:9200,192.168.80.131:9200,192.168.80.132:9200
elasticsearch.connectTimeout=10000
elasticsearch.socketTimeout=60000
elasticsearch.connectionRequestTimeout=10000
elasticsearch.maxConnectNum=200
elasticsearch.maxConnectPerRoute=200
# 无密码可忽略
elasticsearch.userName=elastic
elasticsearch.password=123456
连接配置:
import org.apache.http.HttpHost;
import org.apache.http.auth.AuthScope;
import org.apache.http.auth.UsernamePasswordCredentials;
import org.apache.http.client.CredentialsProvider;
import org.apache.http.impl.client.BasicCredentialsProvider;
import org.elasticsearch.client.RestClient;
import org.elasticsearch.client.RestClientBuilder;
import org.elasticsearch.client.RestHighLevelClient;
import org.springframework.beans.factory.annotation.Value;
import org.springframework.context.annotation.Bean;
import org.springframework.context.annotation.Configuration;
import javax.annotation.PreDestroy;
import java.io.IOException;
import java.util.ArrayList;
import java.util.List;
@Configuration
public class ElasticSearchConfig {
@Value("${elasticsearch.schema:http}")
private String schema;
@Value("${elasticsearch.address}")
private String address;
@Value("${elasticsearch.userName}")
private String userName;
@Value("${elasticsearch.password}")
private String password;
@Value("${elasticsearch.connectTimeout:5000}")
private int connectTimeout;
@Value("${elasticsearch.socketTimeout:10000}")
private int socketTimeout;
@Value("${elasticsearch.connectionRequestTimeout:5000}")
private int connectionRequestTimeout;
@Value("${elasticsearch.maxConnectNum:100}")
private int maxConnectNum;
@Value("${elasticsearch.maxConnectPerRoute:100}")
private int maxConnectPerRoute;
private RestHighLevelClient restHighLevelClient;
@Bean
public RestHighLevelClient restHighLevelClient() {
final CredentialsProvider credentialsProvider = new BasicCredentialsProvider();
UsernamePasswordCredentials elastic = new UsernamePasswordCredentials(userName, password);
credentialsProvider.setCredentials(AuthScope.ANY,elastic);
// 拆分地址
List<HttpHost> hostLists = new ArrayList<>();
String[] hostList = address.split(",");
for (String addr : hostList) {
String host = addr.split(":")[0];
String port = addr.split(":")[1];
hostLists.add(new HttpHost(host, Integer.parseInt(port), schema));
}
// 转换成 HttpHost 数组
HttpHost[] httpHost = hostLists.toArray(new HttpHost[]{});
// 构建连接对象
RestClientBuilder builder = RestClient.builder(httpHost);
// 异步连接延时配置
builder.setRequestConfigCallback(requestConfigBuilder -> {
requestConfigBuilder.setConnectTimeout(connectTimeout);
requestConfigBuilder.setSocketTimeout(socketTimeout);
requestConfigBuilder.setConnectionRequestTimeout(connectionRequestTimeout);
return requestConfigBuilder;
});
// 异步连接数配置
builder.setHttpClientConfigCallback(httpClientBuilder -> {
httpClientBuilder.setMaxConnTotal(maxConnectNum);
httpClientBuilder.setMaxConnPerRoute(maxConnectPerRoute);
httpClientBuilder.setDefaultCredentialsProvider(credentialsProvider);
return httpClientBuilder;
});
restHighLevelClient = new RestHighLevelClient(builder);
return restHighLevelClient;
}
@PreDestroy
public void clientClose() {
try {
this.restHighLevelClient.close();
} catch (IOException e) {
e.printStackTrace();
}
}
}
2、API操作ES
2.1 查询索引列表
可以模糊匹配索引名称
@Test
public void tset() throws IOException {
GetIndexRequest getIndexRequest = new GetIndexRequest("log*");
// 获取es前缀过滤下所有索引
GetIndexResponse getIndexResponse = restHighLevelClient.indices().get(getIndexRequest, RequestOptions.DEFAULT);
// 将es查出的索引转换为list
List<String> elasticsearchList = new ArrayList<>(getIndexResponse.getMappings().keySet());
elasticsearchList.forEach(System.out::println);
}
2.2 TermsQuery
es 的 trem query 做的是精确匹配查询,关于这里早 serviceName 字段后面加的 .keyword 说明如下:
1.es5.0 及以后的版本取消了 String 类型,将原先的 String 类型拆分为 text 和 keyword 两种类型。它们的区别在于 text 会对字段进行分词处理而 keyword 则不会。
2.当没有为索引字段预先指定 mapping 的话,es 就会使用 Dynamic Mapping ,通过推断你传入的文档中字段的值对字段进行动态映射。例如传入的文档中字段 total 的值为12,那么 total 将被映射为 long 类型;字段 addr 的值为"192.168.0.1",那么 addr 将被映射为 ip 类型。然而对于不满足 ip 和 long 格式的普通字符串来说,情况有些不同:ES 会将它们映射为 text 类型,但为了保留对这些字段做精确查询以及聚合的能力,又同时对它们做了 keyword 类型的映射,作为该字段的 fields 属性写到 _mapping 中。例如,我这里使用的字段 “serviceName”,用来存储服务名称字符串类型,会对它做如下的 Dynamic Mapping:
"serviceName" : {
"type" : "text",
"fields" : {
"keyword" : {
"type" : "keyword",
"ignore_above" : 256
}
}
}
在之后的查询中使用 serviceName 是将 serviceName 作为 text 类型查询,而使用 serviceName.keyword 则是将 serviceName 作为 keyword 类型查询。前者会对查询内容做分词处理之后再匹配,而后者则是直接对查询结果做精确匹配。
3.es 的 trem query 做的是精确匹配而不是分词查询,因此对 text 类型的字段做 term 查询将是查不到结果的(除非字段本身经过分词器处理后不变,未被转换或分词)。此时,必须使用 serviceName.keyword 来对 serviceName 字段以 keyword 类型进行精确匹配。
GET logdata-log-center-2021.05.06/_search
{
"query": {
"terms": {
"serviceName.keyword": [
"log-center-user-portal",
"log-center-collect-manage"
]
}
}
}
Java API
@Test
public void test() throws IOException {
//构建查询源构建器
SearchSourceBuilder searchSourceBuilder = new SearchSourceBuilder();
// termQuery只能匹配一个值,第一个入参为字段名称,第二个参数为传入的值,相当于sql中的=
// searchSourceBuilder.query(QueryBuilders.termQuery("serviceName.keyword", "log-center-user-portal-web"));
//termsQuery可以一次性匹配多个值,相当于sql中的in
searchSourceBuilder.query(QueryBuilders.termsQuery("serviceName.keyword", "log-center-user-portal-web", "log-center-collect-manage"));
//构建查询请求对象,入参为索引
SearchRequest searchRequest = new SearchRequest("log-web-up-log-center-2021.10.30");
//向搜索请求对象中配置搜索源
searchRequest.source(searchSourceBuilder);
// 执行搜索,向ES发起http请求
SearchResponse response = restHighLevelClient.search(searchRequest, RequestOptions.DEFAULT);
if (RestStatus.OK.equals(response.status())) {
long total = response.getHits().getTotalHits().value; //检索到符合条件的总数
SearchHit[] hits = response.getHits().getHits();
//未指定size,默认查询的是10条
for (SearchHit hit : hits) {
String index = hit.getIndex();//索引名称
String id = hit.getId(); //文档id
JSONObject jsonObject = JSON.parseObject(hit.getSourceAsString(), JSONObject.class); //文档内容
System.out.println(jsonObject);
}
}
}
2.3 WildcardQuery
es的 wildcard query 做的是模糊匹配查询,类似 sql 中的 like,而 value 值前后的 “*” 号类似与 sql 中的 ”%“ 。
GET logdata-log-center-2021.05.06/_search
{
"query": {
"wildcard": {
"serviceName.keyword": {
"value": "*user-portal*"
}
}
}
}
Java API
searchSourceBuilder.query(QueryBuilders.wildcardQuery("serviceName.keyword", "*" + "user-portal" + "*"));
2.4 RangeQuery
es 的 range query 做的是范围查询,相当于 sql 中的 between … and …
GET log-web-up-log-center-2021.10.30/_search
{
"query": {
"range": {
"timestamp": {
"gte": "2021-10-30 15:00:00",
"lte": "2021-10-30 16:00:00",
"format": "yyyy-MM-dd HH:mm:ss||yyyy-MM-dd HH:mm:ss.SSS"
}
}
}
}
Java API
searchSourceBuilder.query(QueryBuilders.rangeQuery("timestamp")
.gte("2021-10-30 15:00:00") //起始值
.lte("2021-10-30 16:00:00") //结束值
.format("yyyy-MM-dd HH:mm:ss||yyyy-MM-dd HH:mm:ss.SSS"));//可以指定多个格式化标准,使用||隔开
2.5 MatchQuery
es的 match query 做的是全文检索,会对关键字进行分词后匹配词条。
GET log-web-up-log-center-2021.10.30/_search
{
"query": {
"match": {
"orgName": {
"query": "有限公司"
}
}
}
}
query:搜索的关键字,对于英文关键字如果有多个单词则中间要用半角逗号分隔,而对于中文关键字中间可以用逗号分隔也可以不用。
Java API
//全文检索,支持分词匹配
searchSourceBuilder.query(QueryBuilders.matchQuery("orgName", "有限公司");
2.6 MultiMatchQuery
上面的 MatchQuery 有一个短板,假如用户输入了某关键字,我们在检索的时候不知道具体是哪一个字段,这时我们用什么都不合适,而 MultiMatchQuery 的出现解决了这个问题,他可以通过 fields 属性来设置多个域联合查找,具体用法如下
GET log-web-up-log-center-2021.10.30/_search
{
"query": {
"multi_match": {
"query": "user-portal",
"fields": ["serviceName", "systemName"]
}
}
}
Java API
//全文检索,支持分词匹配,支持多字段检索
searchSourceBuilder.query(QueryBuilders.multiMatchQuery("user-portal", "serviceName", "systemName", "description"));
2.7 ExistsQuery
es的 exists query 做的是检索某个字段存在的数据,即不为 null 的数据。其中指定的 field 可以是一个具体的字段,也可以是一个 json 结构。
GET logdata-log-center-2021.05.06/_search
{
"query": {
"exists": {
"field": "networkLogDetailInfo"
}
}
}
Java API
//查询networkLogDetailInfo不为null的数据
searchSourceBuilder.query(QueryBuilders.existsQuery("networkLogDetailInfo"));
2.8 BoolQuery
es的 bool query 做的是将多个查询组合起来去检索数据,主要的组合参数有 must、should、mustNot 等。
must
:数据必须匹配 must 所包含的查询条件,相当于 ”AND“should
:数据匹配 should 包含的一个或多个查询条件,相当于 ”OR“mustNot
:数据必须不匹配 mustNot 所包含的查询条件,相当于 ”NOT“
GET logdata-log-center-2021.05.06/_search
{
"query": {
"bool": {
"must": [
{
"exists": {
"field": "networkLogDetailInfo"
}
},
{
"range": {
"timestamp": {
"gte": "2021-05-05 00:00:00",
"lte": "2021-05-07 00:00:00",
"format": "yyyy-MM-dd HH:mm:ss||yyyy-MM-dd HH:mm:ss.SSS"
}
}
}
],
"must_not": [
{
"exists": {
"field": "serviceLogDetailInfo"
}
}
]
}
}
}
Java API
@Test
public void test() throws IOException {
//构建查询源构建器
SearchSourceBuilder searchSourceBuilder = new SearchSourceBuilder();
//构建bool类型查询器
BoolQueryBuilder boolQueryBuilder = QueryBuilders.boolQuery();
//使用must连接,相当于and,构建第一个查询条件existsQuery必须包含此字段
boolQueryBuilder.must(QueryBuilders.existsQuery("networkLogDetailInfo"));
//使用must连接第二个条件,rangeQuery范围查找,相当于between...and...
boolQueryBuilder.must(QueryBuilders.rangeQuery("timestamp")
.from("2021-05-05 00:00:00") //起始值
.to("2021-05-07 00:00:00") //结束值
.includeLower(true) //是否等于起始值
.includeUpper(false) //是否等于结束值
.format("yyyy-MM-dd HH:mm:ss||yyyy-MM-dd HH:mm:ss.SSS")); //格式化时间
//使用mustNot连接第三个条件
boolQueryBuilder.mustNot(QueryBuilders.existsQuery("serviceLogDetailInfo"));
searchSourceBuilder.query(boolQueryBuilder);
//构建查询请求对象,入参为索引
SearchRequest searchRequest = new SearchRequest("logdata-log-center-2021.05.06");
//向搜索请求对象中配置搜索源
searchRequest.source(searchSourceBuilder);
// 执行搜索,向ES发起http请求
SearchResponse response = restHighLevelClient.search(searchRequest, RequestOptions.DEFAULT);
if (RestStatus.OK.equals(response.status())) {
long total = response.getHits().getTotalHits().value; //检索到符合条件的总数
SearchHit[] hits = response.getHits().getHits();
for (SearchHit hit : hits) {
String index = hit.getIndex();//索引名称
String id = hit.getId(); //文档id
JSONObject jsonObject = JSON.parseObject(hit.getSourceAsString(), JSONObject.class); //文档内容
System.out.println(jsonObject);
}
}
}
2.9 排序
es 使用 sort 进行排序,可以多个字段联合排序。
GET logdata-log-center-2021.05.06/_search
{
"query": {
"bool": {
"must_not": [
{
"exists": {
"field": "serviceLogDetailInfo"
}
}
]
}
},
"sort": [
{
"serviceName.keyword": {
"order": "asc"
},
"timestamp": {
"order": "desc"
}
}
]
}
先按照第一个字段排序,第一个字段相同时按照第二个字段排序。
Java API
//升序
searchSourceBuilder.sort("serviceName.keyword", SortOrder.ASC);
//降序
searchSourceBuilder.sort("timestamp", SortOrder.DESC);
2.10 结果字段过滤
检索数据,有时只需要其中的几个字段,es 也支持对结果集进行字段筛选过滤。字段可以使用 “*” 进行模糊匹配。
GET logdata-log-center-2021.05.06/_search
{
"_source": {
"includes": ["messageId", "system*", "service*", "timestamp"],
"excludes": []
}
}
Java API
//筛选字段,第一个参数为需要的字段,第二个参数为不需要的字段
searchSourceBuilder.fetchSource(new String[] {"messageId", "system*", "service*", "timestamp"}, new String[] {});
2.11 分页
es 的分页方式有三种:from+ size、scroll、search_after, 默认采用的分页方式是 from+ size 的形式。
2.11.1 from+ size
GET logdata-log-center-2021.05.06/_search
{
"from": 0,
"size": 2,
"query": {
"exists": {
"field": "networkLogDetailInfo"
}
},
"_source": {
"includes": ["messageId", "system*", "service*", "timestamp"],
"excludes": []
}
}
通过查询结果可以发现,我们设置了分页参数之后, hits.total 返回的是数据总数7149,而按照分页规则,我们设置的size=2,因此 hits.hits 里面只有两条数据。
Java API
@Test
public void test() throws IOException {
//构建查询源构建器
SearchSourceBuilder searchSourceBuilder = new SearchSourceBuilder();
//查询条件
searchSourceBuilder.query(QueryBuilders.existsQuery("networkLogDetailInfo"));
int page = 1; // 页码
int size = 2; // 每页显示的条数
int index = (page - 1) * size;
searchSourceBuilder.from(index); //设置查询起始位置
searchSourceBuilder.size(size); //结果集返回的数据条数
//筛选字段,第一个参数为需要的字段,第二个参数为不需要的字段
searchSourceBuilder.fetchSource(new String[] {"messageId", "system*", "service*", "timestamp"}, new String[] {});
//构建查询请求对象,入参为索引
SearchRequest searchRequest = new SearchRequest("logdata-log-center-2021.05.06");
//向搜索请求对象中配置搜索源
searchRequest.source(searchSourceBuilder);
// 执行搜索,向ES发起http请求
SearchResponse response = restHighLevelClient.search(searchRequest, RequestOptions.DEFAULT);
if (RestStatus.OK.equals(response.status())) {
long total = response.getHits().getTotalHits().value; //检索到符合条件的总数
SearchHit[] hits = response.getHits().getHits();
//未指定size,默认查询的是10条
for (SearchHit hit : hits) {
String index = hit.getIndex();//索引名称
String id = hit.getId(); //文档id
JSONObject jsonObject = JSON.parseObject(hit.getSourceAsString(), JSONObject.class); //文档内容
System.out.println(jsonObject);
}
}
}
2.11.2 scroll
一种可满足深度分页的方式,es 提供了 scroll 的方式进行分页读取。原理上是对某次查询生成一个游标 scroll_id , 后续的查询只需要根据这个游标去取数据,每次只能拿到下一页的数据,直到结果集中返回的 hits 字段为空,就表示遍历结束。这里scroll=1m是scroll_id的有效期,表示1分钟,过期后会被es自动清理,每次查询会更新此值。
GET logdata-log-center-2021.05.06/_search?scroll=1m
{
"size": 2,
"query": {
"exists": {
"field": "networkLogDetailInfo"
}
},
"_source": {
"includes": ["messageId", "system*", "service*", "timestamp"],
"excludes": []
}
}
后续的查询中查询条件不需要指定,只需要携带 scroll_id 即可它会按照首次查询条件进行分页展示,下一次查询(两种方式):
POST /_search/scroll
{
"scroll": "1m",
"scroll_id": "FGluY2x1ZGVfY29udGV4dF91dWlkDXF1ZXJ5QW5kRmV0Y2gBFFp0bGhXbjBCQU55Q3EtSDcxaWF4AAAAAACF-OYWV0liWUNLUHVTN09DS1ZtUl9SSHhVdw=="
}
GET /_search/scroll?scroll=1m&scroll_id=FGluY2x1ZGVfY29udGV4dF91dWlkDXF1ZXJ5QW5kRmV0Y2gBFFp0bGhXbjBCQU55Q3EtSDcxaWF4AAAAAACF-OYWV0liWUNLUHVTN09DS1ZtUl9SSHhVdw==
Java API
public void testScroll(String scrollId) throws IOException {
//查询源构建器
SearchSourceBuilder searchSourceBuilder = new SearchSourceBuilder();
//每页显示2条
searchSourceBuilder.size(2);
//查询条件
searchSourceBuilder.query(QueryBuilders.existsQuery("networkLogDetailInfo"));
//筛选字段,第一个参数为需要的字段,第二个参数为不需要的字段
searchSourceBuilder.fetchSource(new String[] {"messageId", "system*", "service*", "timestamp"}, new String[] {});
SearchRequest request = new SearchRequest("logdata-log-center-2021.05.06");
request.source(searchSourceBuilder);
Scroll scroll = new Scroll(TimeValue.timeValueMinutes(1L));
request.scroll(scroll);//滚动翻页
SearchResponse response;
if (!StringUtils.isBlank(scrollId)) {
//Scroll查询
SearchScrollRequest scrollRequest = new SearchScrollRequest(scrollId);
scrollRequest.scroll(scroll);
response = restHighLevelClient.scroll(scrollRequest, RequestOptions.DEFAULT);
} else {
//首次查询使用普通查询
response = restHighLevelClient.search(request, RequestOptions.DEFAULT);
}
//更新scrollId
scrollId = response.getScrollId();
System.out.println(scrollId);
if (RestStatus.OK.equals(response.status())) {
//设置查询总量
SearchHit[] hits = response.getHits().getHits();
for (SearchHit hit : hits) {
String index = hit.getIndex();
String id = hit.getId();
JSONObject jsonObject = JSON.parseObject(hit.getSourceAsString(), JSONObject.class);
System.out.println(jsonObject);
}
}
}
2.11.3 search_after
search_after 是 ES5.0 及之后版本提供的新特性,search_after查询时需要指定sort排序字段,可以指定多个排序字段,后续查询有点类似 scroll ,但是和 scroll 又不一样,它提供一个活动的游标,通过上一次查询的最后一条数据的来进行下一次查询。 这里需要说明一下,使用search_after查询需要将from设置为0或-1,当然你也可以不写
第一次查询:
POST logdata-log-center-2021.05.06/_search
{
"size": 2,
"query": {
"exists": {
"field": "networkLogDetailInfo"
}
},
"_source": {
"includes": ["messageId", "system*", "service*", "timestamp"],
"excludes": []
},
"sort": [
{
"timestamp": {
"order": "desc"
}
}
]
}
查询结果:可以看到每一条数据都有一个sort部分,而下一页的查询需要本次查询结果最后一条的sort值作为游标,实现分页查询
第二次查询:
POST logdata-log-center-2021.05.06/_search
{
"search_after": [
1620374316433
],
"size": 2,
"query": {
"exists": {
"field": "networkLogDetailInfo"
}
},
"_source": {
"includes": ["messageId", "system*", "service*", "timestamp"],
"excludes": []
},
"sort": [
{
"timestamp": {
"order": "desc"
}
}
]
}
Java API
public void testSearchAfter(Object[] values) throws IOException {
//查询源构建器
SearchSourceBuilder searchSourceBuilder = new SearchSourceBuilder();
searchSourceBuilder.size(2);
searchSourceBuilder.from(0); //searchAfter需要将from设置为0或-1,当然也可以不写
//查询条件
searchSourceBuilder.query(QueryBuilders.existsQuery("networkLogDetailInfo"));
//筛选字段,第一个参数为需要的字段,第二个参数为不需要的字段
searchSourceBuilder.fetchSource(new String[] {"messageId", "system*", "service*", "timestamp"}, new String[] {});
//以时间戳排序
searchSourceBuilder.sort("timestamp", SortOrder.DESC);
if (values != null)
searchSourceBuilder.searchAfter(values);
SearchRequest request = new SearchRequest("logdata-log-center-2021.05.06");
request.source(searchSourceBuilder);
SearchResponse response = restHighLevelClient.search(request, RequestOptions.DEFAULT);
if (RestStatus.OK.equals(response.status())) {
//设置查询总量
SearchHit[] hits = response.getHits().getHits();
for(int i = 0; i < hits.length; i++) {
String index = hits[i].getIndex();
String id = hits[i].getId();
JSONObject jsonObject = JSON.parseObject(hits[i].getSourceAsString(), JSONObject.class);
System.out.println(jsonObject);
if (i == hits.length-1) {
//最后一条数据的sortValue作为下一次查询的游标值
values = hits[i].getSortValues();
System.out.println(Arrays.toString(values));
}
}
}
}
2.11.4 三种分页方式特点
from+size
比较适合浅分页模式,在深度分页的情况下,这种使用方式效率是非常低的,随着分页页码的不断增大,查询的效率会直线下降。比如from = 5000, size=20, es 需要在各个分片上匹配排序并得到5000*20 条有效数据,然后在结果集中取最后20条。除了效率上的问题,还有一个无法解决的问题是,es 目前支持最大的 skip 值是 max_result_window ,默认为 10000 。也就是当 from + size > max_result_window 时,es 将返回错误。scroll
是一种滚屏形式的分页检索,满足深度分页的场景。查询的时候生成一个游标 scroll_id,有效期内每次返回的值是一样的,后续的查询只需要根据这个游标去取数据即可。scroll查询是很耗性能的方式,scroll_id 的生成可以理解为建立了一个临时的历史快照, 系统会耗费大量的资源来保存一份当前查询结果集映像,并且会占用文件描述符,在此之后的增删改查等操作不会影响到这个快照的结果,因此不建议在实时查询中运用。这种方式往往用于非实时处理大量数据的情况,比如要进行数据迁移或者索引变更之类的。search_after
适用于深度分页+ 排序,分页是根据上一页最后一条数据来定位下一页的位置,所以无法跳页请求,同时在分页请求的过程中,如果有索引数据的增删改,这些变更也会实时的反映到游标上。在选择search_after的排序字段时尽量使用比如文档的id或者时间戳等具有唯一性的字段。search_after 相比 from+size 的浅分页以及 scroll 滚屏查询会有很大的性能提升。
2.22 聚合
es 的 aggs 对数据进行聚合查询统计,查询方式如下:
## 统计各系统一个月的日志采集数量
POST log*/_search
{
"size": 0,
"query": {
"range": {
"timestamp": {
"gte": "2021-10-24 00:00:00",
"lte": "2021-11-24 00:00:00",
"format": "yyyy-MM-dd HH:mm:ss"
}
}
},
"aggs": {
"allLog": {
"terms": {
"field": "systemName.keyword",
"size": 10
}
}
}
}
Java API
@Test
public void test() throws IOException {
//按照systemName字段聚合统计各个系统的日志数量
TermsAggregationBuilder bySystemName = AggregationBuilders.terms("allLog").field("systemName.keyword");
RangeQueryBuilder timestamp = QueryBuilders.rangeQuery("timestamp")
.gte("2021-10-24 00:00:00")
.lte("2021-11-24 00:00:00")
.format("yyyy-MM-dd HH:mm:ss");
//查询源构建器
SearchSourceBuilder searchSourceBuilder = new SearchSourceBuilder();
//配置聚合条件
searchSourceBuilder.aggregation(bySystemName);
//配置查询条件
searchSourceBuilder.query(timestamp);
//设置查询结果不返回,只返回聚合结果
searchSourceBuilder.size(0);
//创建查询请求对象,将查询条件配置到其中
SearchRequest request = new SearchRequest("log*");
request.source(searchSourceBuilder);
// 执行搜索,向ES发起http请求
SearchResponse response = restHighLevelClient.search(request, RequestOptions.DEFAULT);
Aggregations aggregations = response.getAggregations();
if (aggregations != null) {
Terms terms = aggregations.get("allLog");
//解析桶
for (Terms.Bucket bucket : terms.getBuckets()) {
System.out.print("系统名称:" + bucket.getKeyAsString());
System.out.println("\t总日志数量:" + bucket.getDocCount());
}
}
}
多层嵌套聚合
## 统计各个系统的总日志数量,按系统统计各种类型日志数量
POST log*/_search
{
"size": 0,
"query": {
"range": {
"timestamp": {
"gte": "2021-10-24 00:00:00",
"lte": "2021-11-24 00:00:00",
"format": "yyyy-MM-dd HH:mm:ss"
}
}
},
"aggs": {
"allLog": {
"terms": {
"field": "systemName.keyword",
"size": 10
},
"aggs": {
"errorLogNum": {
"filter": {
"terms": {
"level.keyword": [
"ERROR",
"FATAL"
]
}
}
},
"dbLogNum": {
"filter": {
"exists": {
"field": "dataLogDetailInfo"
}
}
},
"interfaceLogNum": {
"filter": {
"exists": {
"field": "networkLogDetailInfo"
}
}
},
"serviceLogNum": {
"filter": {
"exists": {
"field": "serviceLogDetailInfo"
}
}
},
"webLogNum": {
"filter": {
"exists": {
"field": "browserModel"
}
}
}
}
}
}
}
Java API
@Test
public void test() throws IOException {
//错误日志聚合条件
FilterAggregationBuilder errorLogNum = AggregationBuilders.filter("errorLogNum", QueryBuilders.termsQuery("level.keyword", "ERROR", "FATAL"));
//数据库日志聚合条件
FilterAggregationBuilder dataLogNum = AggregationBuilders.filter("dbLogNum", QueryBuilders.existsQuery("dataLogDetailInfo"));
//接口日志聚合条件
FilterAggregationBuilder networkLogNum = AggregationBuilders.filter("interfaceLogNum", QueryBuilders.existsQuery("networkLogDetailInfo"));
//应用日志聚合条件
FilterAggregationBuilder serviceLogNum = AggregationBuilders.filter("serviceLogNum", QueryBuilders.existsQuery("serviceLogDetailInfo"));
//前端日志聚合条件
FilterAggregationBuilder webUpLogNum = AggregationBuilders.filter("webLogNum", QueryBuilders.existsQuery("browserModel"));
//最外层聚合条件,第一次聚合的条件
TermsAggregationBuilder bySystemName = AggregationBuilders.terms("allLog").field("systemName.keyword").size(10);
//内部多个条件的子聚合,在系统聚合后的结果上二次聚合
bySystemName.subAggregation(errorLogNum)
.subAggregation(dataLogNum).
subAggregation(networkLogNum).
subAggregation(serviceLogNum).
subAggregation(webUpLogNum);
RangeQueryBuilder timestamp = QueryBuilders.rangeQuery("timestamp")
.gte("2021-10-24 00:00:00")
.lte("2021-11-24 00:00:00")
.format("yyyy-MM-dd HH:mm:ss");
//查询源构建器
SearchSourceBuilder searchSourceBuilder = new SearchSourceBuilder();
//配置聚合条件
searchSourceBuilder.aggregation(bySystemName);
//配置查询条件
searchSourceBuilder.query(timestamp);
//设置查询结果不返回,只返回聚合结果
searchSourceBuilder.size(0);
//创建查询请求对象,将查询条件配置到其中
SearchRequest request = new SearchRequest("log*");
request.source(searchSourceBuilder);
// 执行搜索,向ES发起http请求
SearchResponse response = restHighLevelClient.search(request, RequestOptions.DEFAULT);
Aggregations aggregations = response.getAggregations();
if (aggregations != null) {
Terms terms = aggregations.get("allLog");
for (Terms.Bucket bucket : terms.getBuckets()) {
ParsedFilter dbFilter = bucket.getAggregations().get("dbLogNum");
ParsedFilter serviceFilter = bucket.getAggregations().get("serviceLogNum");
ParsedFilter webFilter = bucket.getAggregations().get("webLogNum");
ParsedFilter interfaceFilter = bucket.getAggregations().get("interfaceLogNum");
ParsedFilter errorFilter = bucket.getAggregations().get("errorLogNum");
System.out.print("系统名称:" + bucket.getKeyAsString());
System.out.print("\t总日志:" + bucket.getDocCount());
System.out.print("\t数据库日志:" + dbFilter.getDocCount());
System.out.print("\t服务执行日志:" + serviceFilter.getDocCount());
System.out.print("\t前端操作日志:" + webFilter.getDocCount());
System.out.print("\t接口日志:" + interfaceFilter.getDocCount());
System.out.println("\t错误日志:" + errorFilter.getDocCount());
}
}
}
聚合查询还提供了许多查询规则,按时间date聚合、count聚合、avg聚合、sum聚合、min聚合、max聚合等等,这里就不一一列举了。
以上为个人经验,希望能给大家一个参考,也希望大家多多支持编程网。