Centos7安装ElasticSearch 6.4.1入门教程详解
1.下载ElasticSearch 6.4.1安装包 下载地址:
https://artifacts.elastic.co/downloads/elasticsearch/elasticsearch-6.4.1.tar.gz
2.解压压缩包
[root@localhost ElasticSearch]# tar -zxvf elasticsearch-6.4.1.tar.gz
3.启动ElasticSearch
[root@localhost bin]# ./elasticsearch
以后台方式启动
[root@localhost bin]# ./elasticsearch -d
TIPS:
[root@localhost bin]# ./elasticsearch
[2018-09-19T19:46:09,817][WARN ][o.e.b.ElasticsearchUncaughtExceptionHandler] [] uncaught exception in thread [main]
org.elasticsearch.bootstrap.StartupException: java.lang.RuntimeException: can not run elasticsearch as root
at org.elasticsearch.bootstrap.Elasticsearch.init(Elasticsearch.java:140) ~[elasticsearch-6.4.1.jar:6.4.1]
at org.elasticsearch.bootstrap.Elasticsearch.execute(Elasticsearch.java:127) ~[elasticsearch-6.4.1.jar:6.4.1]
at org.elasticsearch.cli.EnvironmentAwareCommand.execute(EnvironmentAwareCommand.java:86) ~[elasticsearch-6.4.1.jar:6.4.1]
at org.elasticsearch.cli.Command.mainWithoutErrorHandling(Command.java:124) ~[elasticsearch-cli-6.4.1.jar:6.4.1]
at org.elasticsearch.cli.Command.main(Command.java:90) ~[elasticsearch-cli-6.4.1.jar:6.4.1]
at org.elasticsearch.bootstrap.Elasticsearch.main(Elasticsearch.java:93) ~[elasticsearch-6.4.1.jar:6.4.1]
at org.elasticsearch.bootstrap.Elasticsearch.main(Elasticsearch.java:86) ~[elasticsearch-6.4.1.jar:6.4.1]
Caused by: java.lang.RuntimeException: can not run elasticsearch as root
at org.elasticsearch.bootstrap.Bootstrap.initializeNatives(Bootstrap.java:104) ~[elasticsearch-6.4.1.jar:6.4.1]
at org.elasticsearch.bootstrap.Bootstrap.setup(Bootstrap.java:171) ~[elasticsearch-6.4.1.jar:6.4.1]
at org.elasticsearch.bootstrap.Bootstrap.init(Bootstrap.java:326) ~[elasticsearch-6.4.1.jar:6.4.1]
at org.elasticsearch.bootstrap.Elasticsearch.init(Elasticsearch.java:136) ~[elasticsearch-6.4.1.jar:6.4.1]
ElasticSearch 不能以root用户角色启动,因此需要将安装目录授权给其他用户,用其他用户来启动
启动成功后,验证,打开新的终端,执行如下命令:
[root@localhost ~]# curl 'http://localhost:9200/?pretty'
{
"name" : "O5BAVYE",
"cluster_name" : "elasticsearch",
"cluster_uuid" : "rw1yjlzkSgODXkUVgIxmxg",
"version" : {
"number" : "6.4.1",
"build_flavor" : "default",
"build_type" : "tar",
"build_hash" : "e36acdb",
"build_date" : "2018-09-13T22:18:07.696808Z",
"build_snapshot" : false,
"lucene_version" : "7.4.0",
"minimum_wire_compatibility_version" : "5.6.0",
"minimum_index_compatibility_version" : "5.0.0"
},
"tagline" : "You Know, for Search"
}
[root@localhost ~]#
返回信息则表示安装成功!
4.安装Kibana
Sense 是一个 Kibana 应用 它提供交互式的控制台,通过你的浏览器直接向 Elasticsearch 提交请求。 这本书的在线版本包含有一个 View in Sense 的链接,里面有许多代码示例。当点击的时候,它会打开一个代码示例的Sense控制台。 你不必安装 Sense,但是它允许你在本地的 Elasticsearch 集群上测试示例代码,从而使本书更具有交互性。
下载kibana
Kibana是一个为 ElasticSearch 提供的数据分析的 Web 接口。可使用它对日志进行高效的搜索、可视化、分析等各种操作
https://artifacts.elastic.co/downloads/kibana/kibana-6.4.1-linux-x86_64.tar.gz
下载完成解压Kibana
[root@localhost ElasticSearch]# tar -zxvf kibana-6.4.1-linux-x86_64.tar.gz
修改 配置config目录下的kibana.yml 文件,配置elasticsearch地址和kibana地址信息
server.host: "192.168.92.50" # kibana 服务器地址
elasticsearch.url: "http://192.168.92.50:9200" # ES 地址
启动 Kibana
[root@localhost bin]# ./kibana
安装Kibana本机访问:http://localhost:5601/
选择Dev Tools菜单,即可实现可视化请求
5.安装LogStash
下载logStash
https://artifacts.elastic.co/downloads/logstash/logstash-7.0.1.tar.gz
下载完成解压后,config目录下配置日志收集日志配置文件 logstash.conf
# Sample Logstash configuration for creating a simple
# Beats -> Logstash -> Elasticsearch pipeline.
input {
tcp {
mode => "server"
host => "192.168.92.50"
port => 4560
codec => json_lines
}
}
output {
elasticsearch {
hosts => "192.168.92.50:9200"
index => "springboot-logstash-%{+YYYY.MM.dd}"
}
}
配置成功后启动logstatsh
[root@localhost bin]# ./logstash -f ../config/logstash.conf
ES 一些基础知识:
索引(名词):
如前所述,一个 索引 类似于传统关系数据库中的一个 数据库 ,是一个存储关系型文档的地方。 索引 (index) 的复数词为 indices 或 indexes 。
索引(动词):
索引一个文档 就是存储一个文档到一个 索引 (名词)中以便它可以被检索和查询到。这非常类似于 SQL 语句中的 INSERT 关键词,除了文档已存在时新文档会替换旧文档情况之外。
倒排索引:
关系型数据库通过增加一个 索引 比如一个 B树(B-tree)索引 到指定的列上,以便提升数据检索速度。Elasticsearch 和 Lucene 使用了一个叫做 倒排索引 的结构来达到相同的目的。
PUT /megacorp/employee/1
{
"first_name" : "John",
"last_name" : "Smith",
"age" : 25,
"about" : "I love to go rock climbing",
"interests": [ "sports", "music" ]
}
返回结果:
#! Deprecation: the default number of shards will change from [5] to [1] in 7.0.0; if you wish to continue using the default of [5] shards, you must manage this on the create index request or with an index template
{
"_index": "megacorp",
"_type": "employee",
"_id": "1",
"_version": 1,
"result": "created",
"_shards": {
"total": 2,
"successful": 1,
"failed": 0
},
"_seq_no": 0,
"_primary_term": 1
}
路径 /megacorp/employee/1 包含了三部分的信息:
megacorp 索引名称
employee 类型名称
1 特定雇员的ID
放置第二个雇员信息:
{
"_index": "megacorp",
"_type": "employee",
"_id": "2",
"_version": 1,
"result": "created",
"_shards": {
"total": 2,
"successful": 1,
"failed": 0
},
"_seq_no": 0,
"_primary_term": 1
}
返回结果:
{
"_index": "megacorp",
"_type": "employee",
"_id": "2",
"_version": 1,
"result": "created",
"_shards": {
"total": 2,
"successful": 1,
"failed": 0
},
"_seq_no": 0,
"_primary_term": 1
}
放置第三个雇员信息
{
"_index": "megacorp",
"_type": "employee",
"_id": "3",
"_version": 1,
"result": "created",
"_shards": {
"total": 2,
"successful": 1,
"failed": 0
},
"_seq_no": 0,
"_primary_term": 1
}
5.检索文档
检索到单个雇员的数据
GET /megacorp/employee/1
返回结果:
{
"_index": "megacorp",
"_type": "employee",
"_id": "1",
"_version": 1,
"found": true,
"_source": {
"first_name": "John",
"last_name": "Smith",
"age": 25,
"about": "I love to go rock climbing",
"interests": [
"sports",
"music"
]
}
}
6.轻量搜索
一个 GET 是相当简单的,可以直接得到指定的文档。 现在尝试点儿稍微高级的功能,比如一个简单的搜索!
第一个尝试的几乎是最简单的搜索了。我们使用下列请求来搜索所有雇员:
GET /megacorp/employee/_search
返回结果:
{
"took": 31,
"timed_out": false,
"_shards": {
"total": 5,
"successful": 5,
"skipped": 0,
"failed": 0
},
"hits": {
"total": 3,
"max_score": 1,
"hits": [
{
"_index": "megacorp",
"_type": "employee",
"_id": "2",
"_score": 1,
"_source": {
"first_name": "Jane",
"last_name": "Smith",
"age": 32,
"about": "I like to collect rock albums",
"interests": [
"music"
]
}
},
{
"_index": "megacorp",
"_type": "employee",
"_id": "1",
"_score": 1,
"_source": {
"first_name": "John",
"last_name": "Smith",
"age": 25,
"about": "I love to go rock climbing",
"interests": [
"sports",
"music"
]
}
},
{
"_index": "megacorp",
"_type": "employee",
"_id": "3",
"_score": 1,
"_source": {
"first_name": "Douglas",
"last_name": "Fir",
"age": 35,
"about": "I like to build cabinets",
"interests": [
"forestry"
]
}
}
]
}
}
通过姓名模糊匹配来获得结果
GET /megacorp/employee/_search?q=last_name:Smith
返回结果:
{
"took": 414,
"timed_out": false,
"_shards": {
"total": 5,
"successful": 5,
"skipped": 0,
"failed": 0
},
"hits": {
"total": 2,
"max_score": 0.2876821,
"hits": [
{
"_index": "megacorp",
"_type": "employee",
"_id": "2",
"_score": 0.2876821,
"_source": {
"first_name": "Jane",
"last_name": "Smith",
"age": 32,
"about": "I like to collect rock albums",
"interests": [
"music"
]
}
},
{
"_index": "megacorp",
"_type": "employee",
"_id": "1",
"_score": 0.2876821,
"_source": {
"first_name": "John",
"last_name": "Smith",
"age": 25,
"about": "I love to go rock climbing",
"interests": [
"sports",
"music"
]
}
}
]
}
}
7.使用查询表达式搜索
领域特定语言 (DSL), 指定了使用一个 JSON 请求
GET /megacorp/employee/_search
{
"query" : {
"match" : {
"last_name" : "Smith"
}
}
}
返回结果:
{
"took": 7,
"timed_out": false,
"_shards": {
"total": 5,
"successful": 5,
"skipped": 0,
"failed": 0
},
"hits": {
"total": 2,
"max_score": 0.2876821,
"hits": [
{
"_index": "megacorp",
"_type": "employee",
"_id": "2",
"_score": 0.2876821,
"_source": {
"first_name": "Jane",
"last_name": "Smith",
"age": 32,
"about": "I like to collect rock albums",
"interests": [
"music"
]
}
},
{
"_index": "megacorp",
"_type": "employee",
"_id": "1",
"_score": 0.2876821,
"_source": {
"first_name": "John",
"last_name": "Smith",
"age": 25,
"about": "I love to go rock climbing",
"interests": [
"sports",
"music"
]
}
}
]
}
}
8.更复杂的搜索
搜索姓氏为 Smith 的雇员,但这次我们只需要年龄大于 30 的,使用过滤器 filter ,它支持高效地执行一个结构化查询
GET /megacorp/employee/_search
{
"query" : {
"bool": {
"must": {
"match" : {
"last_name" : "smith"
}
},
"filter": {
"range" : {
"age" : { "gt" : 30 }
}
}
}
}
}
其中:range 过滤器 , 它能找到年龄大于 30 的文档,其中 gt 表示_大于(_great than)
返回结果:
{
"took": 44,
"timed_out": false,
"_shards": {
"total": 5,
"successful": 5,
"skipped": 0,
"failed": 0
},
"hits": {
"total": 1,
"max_score": 0.2876821,
"hits": [
{
"_index": "megacorp",
"_type": "employee",
"_id": "2",
"_score": 0.2876821,
"_source": {
"first_name": "Jane",
"last_name": "Smith",
"age": 32,
"about": "I like to collect rock albums",
"interests": [
"music"
]
}
}
]
}
}
9.全文搜索
搜索下所有喜欢攀岩(rock climbing)的雇员
GET /megacorp/employee/_search
{
"query" : {
"match" : {
"about" : "rock climbing"
}
}
}
返回结果:
{
"took": 17,
"timed_out": false,
"_shards": {
"total": 5,
"successful": 5,
"skipped": 0,
"failed": 0
},
"hits": {
"total": 2,
"max_score": 0.5753642,
"hits": [
{
"_index": "megacorp",
"_type": "employee",
"_id": "1",
"_score": 0.5753642,
"_source": {
"first_name": "John",
"last_name": "Smith",
"age": 25,
"about": "I love to go rock climbing",
"interests": [
"sports",
"music"
]
}
},
{
"_index": "megacorp",
"_type": "employee",
"_id": "2",
"_score": 0.2876821,
"_source": {
"first_name": "Jane",
"last_name": "Smith",
"age": 32,
"about": "I like to collect rock albums",
"interests": [
"music"
]
}
}
]
}
}
10.全文搜索
找出一个属性中的独立单词是没有问题的,但有时候想要精确匹配一系列单词或者短语 。 比如, 我们想执行这样一个查询,仅匹配同时包含 “rock” 和 “climbing” ,并且 二者以短语 “rock climbing” 的形式紧挨着的雇员记录。
GET /megacorp/employee/_search
{
"query" : {
"match_phrase" : {
"about" : "rock climbing"
}
}
}
返回结果:
{
"took": 142,
"timed_out": false,
"_shards": {
"total": 5,
"successful": 5,
"skipped": 0,
"failed": 0
},
"hits": {
"total": 1,
"max_score": 0.5753642,
"hits": [
{
"_index": "megacorp",
"_type": "employee",
"_id": "1",
"_score": 0.5753642,
"_source": {
"first_name": "John",
"last_name": "Smith",
"age": 25,
"about": "I love to go rock climbing",
"interests": [
"sports",
"music"
]
}
}
]
}
}
11.高亮搜索
许多应用都倾向于在每个搜索结果中 高亮 部分文本片段,以便让用户知道为何该文档符合查询条件。在 Elasticsearch 中检索出高亮片段也很容易。
增加参数: highlight
GET /megacorp/employee/_search
{
"query" : {
"match_phrase" : {
"about" : "rock climbing"
}
},
"highlight": {
"fields" : {
"about" : {}
}
}
}
返回结果:
{
"took": 250,
"timed_out": false,
"_shards": {
"total": 5,
"successful": 5,
"skipped": 0,
"failed": 0
},
"hits": {
"total": 1,
"max_score": 0.5753642,
"hits": [
{
"_index": "megacorp",
"_type": "employee",
"_id": "1",
"_score": 0.5753642,
"_source": {
"first_name": "John",
"last_name": "Smith",
"age": 25,
"about": "I love to go rock climbing",
"interests": [
"sports",
"music"
]
},
"highlight": {
"about": [
"I love to go <em>rock</em> <em>climbing</em>"
]
}
}
]
}
}
其中高亮模块为highlight属性
12.分析
Elasticsearch 有一个功能叫聚合(aggregations),允许我们基于数据生成一些精细的分析结果。聚合与 SQL 中的 GROUP BY 类似但更强大。
举个例子,挖掘出雇员中最受欢迎的兴趣爱好:
GET /megacorp/employee/_search
{
"aggs": {
"all_interests": {
"terms": { "field": "interests" }
}
}
}
返回结果:
{
...
"hits": { ... },
"aggregations": {
"all_interests": {
"buckets": [
{
"key": "music",
"doc_count": 2
},
{
"key": "forestry",
"doc_count": 1
},
{
"key": "sports",
"doc_count": 1
}
]
}
}
}
以上就是本文的全部内容,希望对大家的学习有所帮助,也希望大家多多支持我们。
免责声明:
① 本站未注明“稿件来源”的信息均来自网络整理。其文字、图片和音视频稿件的所属权归原作者所有。本站收集整理出于非商业性的教育和科研之目的,并不意味着本站赞同其观点或证实其内容的真实性。仅作为临时的测试数据,供内部测试之用。本站并未授权任何人以任何方式主动获取本站任何信息。
② 本站未注明“稿件来源”的临时测试数据将在测试完成后最终做删除处理。有问题或投稿请发送至: 邮箱/279061341@qq.com QQ/279061341