ELK+kafka集成

一、由於本項目採用的log4j2，因此在log4j2中直接配置 

<Kafka name="Kafka" topic="XX_log">  
  <PatternLayout pattern="%d{yyyy-MM-dd HH:mm:ss}||%p||%c{1}||XX_web||%m%n"/>  
    <Property name="bootstrap.servers">127.0.0.1:9092</Property>  
    <Property name="timeout.ms">500</Property>  
</Kafka>

PatternLayout 中格式採用了||將內容鏈接起來目的爲了logstash進行切分，其中增長timeout.ms屬性爲了保證日誌系統掛掉的狀況不會對業務系統產生較大影響，固然kafka能夠採用集羣的方式，bootstrap.servers多個地址用「,」分隔。XX_web表明當前業務平臺。 

 

二、搭建kafka集羣這裏就很少介紹了官網很全, 

zookeeper.connect=127.0.0.1:2181,127.0.0.1:2182,127.0.0.1:2183

 

三、建立logstash動態模板 

{  
    "template": "*",  
    "settings": {  
        "index.refresh_interval": "5s",  
        "number_of_replicas": "0",  
        "number_of_shards": "3"  
    },  
    "mappings": {  
        "_default_": {  
            "_all": {  
                "enabled": false  
            },  
            "dynamic_templates": [  
                {  
                    "message_field": {  
                        "match": "message",  
                        "match_mapping_type": "string",  
                        "mapping": {  
                            "type": "string",  
                            "index": "analyzed"  
                        }  
                    }  
                },  
                {  
                    "string_fields": {  
                        "match": "*",  
                        "match_mapping_type": "string",  
                        "mapping": {  
                            "type": "string",  
                            "index": "not_analyzed"  
                        }  
                    }  
                }  
            ],  
            "properties": {  
                "dateTime": {  
                    "type": "date",  
                    "format": "yyy-MM-dd HH:mm:ss"  
                },  
                "@version": {  
                    "type": "integer",  
                    "index": "not_analyzed"  
                },  
                "context": {  
                    "type": "string",  
                    "index": "analyzed"  
                },  
                "level": {  
                    "type": "string",  
                    "index": "not_analyzed"  
                },  
                "class": {  
                    "type": "string",  
                    "index": "not_analyzed"  
                },  
                "server": {  
                    "type": "string",  
                    "index": "not_analyzed"  
                }  
            }  
        }  
    }  
}

 

四、配置logstash 

input{  
       kafka {  
                zk_connect =>"127.0.0.1:2181,127.0.0.1:2182,127.0.0.1:2183"  
                group_id =>"logstash"  
                topic_id =>"XX_log"  
                reset_beginning => false  
                consumer_threads => 5  
                decorate_events => true  
       }  
}  
filter {  
   mutate{  
        split=>["message","||"]  
        add_field => {  
             "dateTime" => "%{[message][0]}"  
        }  
        add_field => {  
              "level" => "%{[message][1]}"  
        }  
        add_field => {  
               "class" => "%{[message][2]}"  
        }  
        add_field => {  
                "server" => "%{[message][3]}"  
         }  
        add_field => {  
                "context" => "%{[message][4]}"  
         }  
         remove_field => ["message"]  
    }  
    date {  
        match => ["logdate", "yyyy-MM-dd HH:mm:ss"]  
    }  
}  
output{  
      elasticsearch {  
            hosts => ["127.0.0.1:9200"]  
            index => "XX_log-%{+YYYY-MM}"  
            codec => "json"  
            manage_template => true  
            template_overwrite => true  
            flush_size => 50000  
            idle_flush_time => 10  
            workers => 2  
            template => "E:\logstash\template\template_log.json"    
    }  
}

 

按照年月將日誌保存進ES索引中index => "XX_log-%{+YYYY-MM}"，logstash從kafka集羣中讀取日誌信息。 

 

五、搭建ZK集羣，這裏就很少介紹了，網上資料比較多

 

六、搭建ES集羣，ES集羣比較簡單，設置的參數不要太多就可使用。

 

七、配置kibana 

server.port: 5601 # 服務端口  
# The host to bind the server to.  
server.host: "115.28.240.113"  
elasticsearch.url: http://127.0.0.1:9200   ES地址-集羣  
kibana.index: "kibana"

 

八、版本 JKD 1.7  ES-2.4, logstash 2.4, kafka-2.10,kibana-4.6.4