一、下载 从官网下载:https://www.elastic.co/downloads/logstash ; 二、配置 1、修改config下的jvm.options: 1️⃣根据需要修改最大堆和最小堆 2️⃣我这里使用的jdk1.8,gc使用G1,所以需要重新配置; -XX:+UnlockDiagnosticVMOptions -XX:+UseCompressedOops -XX:+UseG1GC -XX:MaxGCPauseMillis=200 -XX:G1ReservePercent=20 -XX:+G1SummarizeConcMark -XX:InitiatingHeapOccupancyPercent=40 -XX:+AlwaysPreTouch -XX:+DisableExplicitGC
2、新建配置文件service.conf
3、在service.conf配置数据收集、解析及输出:
#==数据源配置,从kafka
input{
kafka{
bootstrap_servers => ["10.31.140.96:9092,10.31.140.99:9092,10.31.140.93:9092"]
topics => "elk-service-log"
group_id => "service-log-group"
codec => "json"
auto_offset_reset => "earliest"
max_partition_fetch_bytes => "52428700"
max_poll_records => "300"
consumer_threads => "16"
}
} #==数据解析——这里的日志内容是json格式,所以用以下解析 filter{ mutate{ gsub => ["message", "[\n|\r|\f|\t]", " "] gsub => ["message", "[\]", "/"] } json{ source => "message" #==filebeat收集的日志内容在message字段中,JSON格式 #target => "doc" }
if [type] == "nginx-log" { #==匹配字段内容,去除filebeat中不需要的内容
grok{
match => {
"@version" => "1"
}
remove_field => ["@version"]
remove_field => ["offset"]
remove_field => ["beat"]
remove_field => ["@version"]
remove_field => ["source"]
remove_field => ["input_type"]
remove_field => ["message"]
#remove_field => ["request_body"]
}
} else {
grok{
match => {
"@version" => "1"
}
remove_field => ["@version"]
remove_field => ["offset"]
remove_field => ["beat"]
remove_field => ["@version"]
remove_field => ["source"]
remove_field => ["input_type"]
}
}
} #== 数据输出到elasticsearch集群 output{ if [type] == "service-log" { #==type字段是在filebeat中添加的自定义字段,用于日志内容区分 elasticsearch{ hosts => ["es1:9200","es2:9200","es3:9200"] index => "service-log-%{+yyyy.MM.dd}" #==每天一个日志索引 codec => "json" manage_template => true template => "/mnt/logstash-5.6.4/templates/service-log.json" #日志索引模板 template_name=>"service-log" template_overwrite=>true } }
if [type] == "nginx-log" {
elasticsearch{
hosts => ["es1:9200","es2:9200","es3:9200"]
index => "nginx-log-%{+yyyy.MM.dd}"
codec => "json"
manage_template => true
template => "/mnt/logstash-5.6.4/templates/nginx-log.json"
template_name=>"nginx-log"
template_overwrite=>true
}
}
} 三、日志索引模板: 日志索引模板是用来创建索引时规定字段类型、设置索引配置的,设置得当可以提高Elasticsearch性能,减少Elasticsearch对资源的消耗。 日志模板如下: { "template": "service-log*", "settings": { "index.number_of_shards": 12, "index.number_of_replicas": 0, "index.refresh_interval":"10s" }, "mappings": { "java": { "_all": { "analyzer": "ik_max_word", #==ik分词 "enabled": false #==禁用all }, "properties": { "@timestamp": { "format": "dateOptionalTime", "type": "date" }, "date": { "type": "keyword" }, "tranceId": { "type": "keyword" }, "sequenceId": { "type": "keyword" }, "level": { "type": "keyword" }, "appName": { "type": "text", "analyzer":"ik_max_word" }, "serverName": { "type": "text", "analyzer":"ik_max_word" }, "port": { "type": "integer" }, "class": { "analyzer": "ik_max_word", "type":"text" }, "method": { "type": "text", "analyzer":"ik_max_word" }, "line": { "type": "integer" }, "message": { "analyzer": "ik_max_word", "type": "text" } } } } } 四、启动: bin/logstash -f service.conf &
五、维护: 如果不同的日志内容格式需要接入到ELK,那么在filebeat端增加一个type标识内容,在logstash中根据type进行处理,然后写入Elasticsearch时新建一个索引模板写入即可。
注意事项: template的命名一定要和logstash中输出到Elasticsearch时的配置匹配,否则会导致template不生效,但是还会写入到Elasticsearch; 如:"template": "service-log*",那么配置为 index => "service-log-%{+yyyy.MM.dd}",配置为index => "nginx-log-%{+yyyy.MM.dd}"就会提示找不到template