京东物流:康睿 姚再毅 李振 刘斌 王北永
说明:以下全部均基于 ElasticSearch 8.1 版本
一 跨集群检索 - ccr
官网文档地址:
跨集群检索的背景和意义
跨集群检索定义
跨集群检索环境搭建
官网文档地址:
步骤 1:搭建两个本地单节点集群,本地练习可取消安全配置
步骤 2:每个集群都执行以下命令
PUT _cluster/settings { "persistent": { "cluster": { "remote": { "cluster_one": { "seeds": [ "172.21.0.14:9301" ] },"cluster_two": { "seeds": [ "172.21.0.14:9302" ] } } } } }
步骤 3:验证集群之间是否互通
方案 1:Kibana 可视化查看:stack Management -> Remote Clusters -> status 应该是 connected! 且必须打上绿色的对号。
方案 2:GET _remote/info
跨集群查询演练
# 步骤1 在集群 1 中添加数据如下 PUT test01/_bulk
{"index":{"_id":1}}
{"title":"this is from cluster01..."} # 步骤2 在集群 2 中添加数据如下: PUT test01/_bulk
{"index":{"_id":1}}
{"title":"this is from cluster02..."} # 步骤 3:执行跨集群检索如下: 语法:POST 集群名称1:索引名称,集群名称2:索引名称/_search POST cluster_one:test01,cluster_two:test01/_search
{ "took" : 7, "timed_out" : false, "num_reduce_phases" : 3, "_shards" : { "total" : 2, "successful" : 2, "skipped" : 0, "failed" : 0
}, "_clusters" : { "total" : 2, "successful" : 2, "skipped" : 0
}, "hits" : { "total" : { "value" : 2, "relation" : "eq" }, "max_score" : 1.0, "hits" : [
{ "_index" : "cluster_two:test01", "_id" : "1", "_score" : 1.0, "_source" : { "title" : "this is from cluster02..." }
},
{ "_index" : "cluster_one:test01", "_id" : "1", "_score" : 1.0, "_source" : { "title" : "this is from cluster01..." }
}
]
}
}
二 跨集群复制 - ccs - 该功能需付费
官网文档地址:
https:///guide/en/elasticsearch/reference/current/xpack-ccr.html
如何保障集群的高可用
- 副本机制
- 快照和恢复
- 跨集群复制(类似 mysql 主从同步)
跨集群复制概述
跨集群复制配置
- 准备两个集群,网络互通
- 开启 license 使用,可试用 30 天
- 开启位置:Stack Management -> License mangement.
3. 定义好谁是 Leads 集群,谁是 follower 集群
4. 在 follower 集群配置 Leader 集群
5. 在 follower 集群配置 Leader 集群的索引同步规则(kibana 页面配置)
a.stack Management -> Cross Cluster Replication -> create a follower index.
6. 启用步骤 5 的配置
三索引模板
官网文档地址:
8.X 之组件模板
1. 创建组件模板 - 索引 setting 相关
# 组件模板 - 索引setting相关 PUT _component_template/template_sttting_part { "template": { "settings": { "number_of_shards": 3, "number_of_replicas": 0 } } }
2. 创建组件模板 - 索引 mapping 相关
# 组件模板 - 索引mapping相关 PUT _component_template/template_mapping_part
{ "template": { "mappings": { "properties": { "hosr_name":{ "type": "keyword" }, "cratet_at":{ "type": "date", "format": "EEE MMM dd HH:mm:ss Z yyyy" }
}
}
}
}
3. 创建组件模板 - 配置模板和索引之间的关联
// **注意:composed_of 如果多个组件模板中的配置项有重复,后面的会覆盖前面的,和配置的顺序有关** # 基于组件模板,配置模板和索引之间的关联 # 也就是所有 tem_* 该表达式相关的索引创建时,都会使用到以下规则 PUT _index_template/template_1
{ "index_patterns": [ "tem_*" ], "composed_of": [ "template_sttting_part", "template_mapping_part" ]
}
4. 测试
# 创建测试 PUT tem_001
索引模板基本操作
实战演练
需求 1:默认如果不显式指定 Mapping, 数值类型会被动态映射为 long 类型,但实际上业务数值都比较小,会存在存储浪费。需要将默认值指定为 Integer
索引模板,官网文档地址:
mapping - 动态模板,官网文档地址:
# 结合mapping 动态模板 和 索引模板 # 1.创建组件模板之 - mapping模板 PUT _component_template/template_mapping_part_01
{ "template": { "mappings": { "dynamic_templates": [
{ "integers": { "match_mapping_type": "long", "mapping": { "type": "integer" }
}
}
]
}
}
} # 2. 创建组件模板与索引关联配置 PUT _index_template/template_2
{ "index_patterns": ["tem1_*"], "composed_of": ["template_mapping_part_01"]
} # 3.创建测试数据 POST tem1_001/_doc/1 { "age":18 } # 4.查看mapping结构验证 get tem1_001/_mapping
需求 2:date_* 开头的字段,统一匹配为 date 日期类型。
索引模板,官网文档地址:
mapping - 动态模板,官网文档地址:
# 结合mapping 动态模板 和 索引模板 # 1.创建组件模板之 - mapping模板 PUT _component_template/template_mapping_part_01
{ "template": { "mappings": { "dynamic_templates": [
{ "integers": { "match_mapping_type": "long", "mapping": { "type": "integer" }
}
},
{ "date_type_process": { "match": "date_*", "mapping": { "type": "date", "format":"yyyy-MM-dd HH:mm:ss" }
}
}
]
}
}
} # 2. 创建组件模板与索引关联配置 PUT _index_template/template_2
{ "index_patterns": ["tem1_*"], "composed_of": ["template_mapping_part_01"]
} # 3.创建测试数据 POST tem1_001/_doc/2 { "age":19, "date_aoe":"2022-01-01 18:18:00" } # 4.查看mapping结构验证 get tem1_001/_mapping
四.LIM 索引生命周期管理
官网文档地址:
什么是索引生命周期
索引的 生 -> 老 -> 病 -> 死
是否有过考虑,如果一个索引,创建之后,就不再去管理了?会发生什么?
什么是索引生命周期管理
索引太大了会如何?
大索引的恢复时间,要远比小索引恢复慢的多的多索引大了以后,检索会很慢,写入和更新也会受到不同程度的影响索引大到一定程度,当索引出现健康问题,会导致整个集群核心业务不可用
{BANNED}最佳佳实践
集群的单个分片{BANNED}最佳大文档数上限:2 的 32 次幂减 1,即 20 亿左右官方建议:分片大小控制在 30GB-50GB,若索引数据量无限增大,肯定会超过这个值
用户不关注全量
某些业务场景,业务更关注近期的数据,如近 3 天、近 7 天大索引会将全部历史数据汇集在一起,不利于这种场景的查询
索引生命周期管理的历史演变
LIM 前奏 - rollover 滚动索引
官网文档地址:
# 0.自测前提,lim生命周期rollover频率。默认10分钟 PUT _cluster/settings
{ "persistent": { "indices.lifecycle.poll_interval": "1s" }
} # 1. 创建索引,并指定别名 PUT test_index-0001 { "aliases": { "my-test-index-alias": { "is_write_index": true
}
}
} # 2.批量导入数据 PUT my-test-index-alias/_bulk
{"index":{"_id":1}}
{"title":"testing 01"}
{"index":{"_id":2}}
{"title":"testing 02"}
{"index":{"_id":3}}
{"title":"testing 03"}
{"index":{"_id":4}}
{"title":"testing 04"}
{"index":{"_id":5}}
{"title":"testing 05"} # 3.rollover 滚动规则配置 POST my-test-index-alias/_rollover
{ "conditions": { "max_age": "7d", "max_docs": 5, "max_primary_shard_size": "50gb" }
} # 4.在满足条件的前提下创建滚动索引 PUT my-test-index-alias/_bulk
{"index":{"_id":7}}
{"title":"testing 07"} # 5.查询验证滚动是否成功 POST my-test-index-alias/_search
LIM 前奏 - shrink 索引压缩
官网文档地址:
核心步骤:
1. 将数据全部迁移至一个独立的节点
2. 索引禁止写入
3. 方可进行压缩
# 1.准备测试数据 DELETE kibana_sample_data_logs_ext
PUT kibana_sample_data_logs_ext
{ "settings": { "number_of_shards": 5, "number_of_replicas": 0 }
}
POST _reindex
{ "source": { "index": "kibana_sample_data_logs" }, "dest": { "index": "kibana_sample_data_logs_ext" }
} # 2.压缩前必要的条件设置 # number_of_replicas :压缩后副本为0 # index.routing.allocation.include._tier_preference 数据分片全部路由到hot节点 # "index.blocks.write 压缩后索引不再允许数据写入 PUT kibana_sample_data_logs_ext/_settings
{ "settings": { "index.number_of_replicas": 0, "index.routing.allocation.include._tier_preference": "data_hot", "index.blocks.write": true }
} # 3.实施压缩 POST kibana_sample_data_logs_ext/_shrink/kibana_sample_data_logs_ext_shrink
{ "settings":{ "index.number_of_replicas": 0, "index.number_of_shards": 1, "index.codec":"best_compression" }, "aliases":{ "kibana_sample_data_logs_alias":{}
}
}
LIM 实战
全局认知建立 - 四大阶段
官网文档地址:
生命周期管理阶段(Policy):
Hot 阶段 (生)
Set priority
Unfollow
Rollover
Read-only
Shrink
Force Merge
Search snapshot
Warm 阶段(老)
Set priority
Unfollow
Read-only
Allocate
migrate
Shirink
Force Merge
Cold 阶段(病)
Search snapshot
Delete 阶段(死)
delete
演练
1. 创建 policy
-
Hot 阶段设置,rollover: max_age:3d,max_docs:5, max_size:50gb, 优先级:100
-
Warm 阶段设置:min_age:15s , forcemerage 段合并,热节点迁移到 warm 节点,副本数设置 0,优先级:50
-
Cold 阶段设置: min_age 30s, warm 迁移到 cold 阶段
-
Delete 阶段设置:min_age 45s,执行删除操作
PUT _ilm/policy/kr_20221114_policy
{ "policy": { "phases": { "hot": { "min_age": "0ms", "actions": { "set_priority": { "priority": 100 }, "rollover": { "max_size": "50gb", "max_primary_shard_size": "50gb", "max_age": "3d", "max_docs": 5 }
}
}, "warm": { "min_age": "15s", "actions": { "forcemerge": { "max_num_segments": 1 }, "set_priority": { "priority": 50 }, "allocate": { "number_of_replicas": 0 }
}
}, "cold": { "min_age": "30s", "actions": { "set_priority": { "priority": 0 }
}
}, "delete": { "min_age": "45s", "actions": { "delete": { "delete_searchable_snapshot": true }
}
}
}
}
}
2. 创建 index template
PUT _index_template/kr_20221114_template
{ "index_patterns": ["kr_index-**"], "template": { "settings": { "index": { "lifecycle": { "name": "kr_20221114_policy", "rollover_alias": "kr-index-alias" }, "routing": { "allocation": { "include": { "_tier_preference": "data-hot" }
}
}, "number_of_shards": "3", "number_of_replicas": "1" }
}, "aliases": {}, "mappings": {}
}
}
3. 测试需要修改 lim rollover 刷新频率
PUT _cluster/settings
{ "persistent": { "indices.lifecycle.poll_interval": "1s" }
}
4. 进行测试
# 创建索引,并制定可写别名 PUT kr_index-0001 { "aliases": { "kr-index-alias": { "is_write_index": true }
}
} # 通过别名新增数据 PUT kr-index-alias/_bulk
{"index":{"_id":1}}
{"title":"testing 01"}
{"index":{"_id":2}}
{"title":"testing 02"}
{"index":{"_id":3}}
{"title":"testing 03"}
{"index":{"_id":4}}
{"title":"testing 04"}
{"index":{"_id":5}}
{"title":"testing 05"} # 通过别名新增数据,触发rollover PUT kr-index-alias/_bulk
{"index":{"_id":6}}
{"title":"testing 06"} # 查看索引情况 GET kr_index-0001 get _cat/indices?v
过程总结
{BANNED}中国第一步:配置 lim pollicy
-
横向:Phrase 阶段 (Hot、Warm、Cold、Delete) 生老病死
-
纵向:Action 操作(rollover、forcemerge、readlyonly、delete)
第二步:创建模板 绑定 policy, 指定别名
第三步:创建起始索引
第四步:索引基于{BANNED}中国第一步指定的 policy 进行滚动
五.Data Stream
官网文档地址:
特性解析
Data Stream 让我们跨多个索引存储时序数据,同时给了唯一的对外接口(data stream 名称)
-
写入和检索请求发给 data stream
-
data stream 将这些请求路由至 backing index(后台索引)
Backing indices
每个 data stream 由多个隐藏的后台索引构成
-
自动创建
-
要求模板索引
rollover 滚动索引机制用于自动生成后台索引
- 将成为 data stream 新的写入索引
应用场景
- 日志、事件、指标等其他持续创建(少更新)的业务数据
- 两大核心特点
- 时序性数据
- 数据极少更新或没有更新
创建 Data Stream 核心步骤
官网文档地址:
Set up a data stream
To set up a data stream, follow these steps:
- Create an index lifecycle policy
- Create component templates
- Create an index template
- Create the data stream
- Secure the data stream
演练
1. 创建一个 data stream,名称为 my-data-stream
2. index_template 名称为 my-index-template
3. 满足 index 格式【"my-data-stream*"】的索引都要被应用到
4. 数据插入的时候,在 data_hot 节点
5. 过 3 分钟之后要 rollover 到 data_warm 节点
6. 再过 5 分钟要到 data_cold 节点
# 步骤1 。创建 lim policy PUT _ilm/policy/my-lifecycle-policy
{ "policy": { "phases": { "hot": { "actions": { "rollover": { "max_size": "50gb", "max_age": "3m", "max_docs": 5 }, "set_priority": { "priority": 100 }
}
}, "warm": { "min_age": "5m", "actions": { "allocate": { "number_of_replicas": 0 }, "forcemerge": { "max_num_segments": 1 }, "set_priority": { "priority": 50 }
}
}, "cold": { "min_age": "6m", "actions": { "freeze":{}
}
}, "delete": { "min_age": "45s", "actions": { "delete": {}
}
}
}
}
} # 步骤2 创建组件模板 - mapping PUT _component_template/my-mappings
{ "template": { "mappings": { "properties": { "@timestamp": { "type": "date", "format": "date_optional_time||epoch_millis" }, "message": { "type": "wildcard" }
}
}
}, "_meta": { "description": "Mappings for @timestamp and message fields", "my-custom-meta-field": "More arbitrary metadata" }
} # 步骤3 创建组件模板 - setting PUT _component_template/my-settings
{ "template": { "settings": { "index.lifecycle.name": "my-lifecycle-policy", "index.routing.allocation.include._tier_preference":"data_hot" }
}, "_meta": { "description": "Settings for ILM", "my-custom-meta-field": "More arbitrary metadata" }
} # 步骤4 创建索引模板 PUT _index_template/my-index-template
{ "index_patterns": ["my-data-stream*"], "data_stream": { }, "composed_of": [ "my-mappings", "my-settings" ], "priority": 500, "_meta": { "description": "Template for my time series data", "my-custom-meta-field": "More arbitrary metadata" }
} # 步骤5 创建 data stream 并 写入数据测试 PUT my-data-stream/_bulk
{ "create":{ } }
{ "@timestamp": "2099-05-06T16:21:15.000Z", "message": "192.0.2.42 - - [06/May/2099:16:21:15 +0000] \"GET /images/bg.jpg HTTP/1.0\" 200 24736" }
{ "create":{ } }
{ "@timestamp": "2099-05-06T16:25:42.000Z", "message": "192.0.2.255 - - [06/May/2099:16:25:42 +0000] \"GET /favicon.ico HTTP/1.0\" 200 3638" }
POST my-data-stream/_doc
{ "@timestamp": "2099-05-06T16:21:15.000Z", "message": "192.0.2.42 - - [06/May/2099:16:21:15 +0000] \"GET /images/bg.jpg HTTP/1.0\" 200 24736" } # 步骤6 查看data stream 后台索引信息 GET /_resolve/index/my-data-stream*