1.从官网下载安装包，并通过Xftp5上传到机器集群上
下载logstash-6.2.3.tar.gz版本，并通过Xftp5上传到hadoop机器集群的第一个节点node1上的/opt/uploads/目录：
2、解压logstash-6.2.3.tar.gz，并把解压的安装包移动到/opt/app/目录上
tar zxvf logstash-6.2.3.tar.gz
mv logstash-6.2.3 /opt/app/ && cd /opt/app/
3、修改环境变量，编辑/etc/profile，并生效环境变量，输入如下命令：
sudo vi /etc/profile
添加如下内容：
export LOGSTASH_HOME=/opt/app/logstash-6.2.3
export PATH=:$PATH:$LOGSTASH_HOME/bin
使环境变量生效：source /etc/profile
4、配置文件类型
4.1 log-kafka配置文件
输入源为nginx的日志文件，输出源为kafka
input {
file {
path => "/var/logs/nginx/*.log"
discover_interval => 5
start_position => "beginning"
}
}
output {
kafka {
topic_id => "accesslog"
codec => plain {
format => "%{message}"
charset => "UTF-8"
}
bootstrap_servers => "hadoop1:9092,hadoop2:9092,hadoop3:9092"
}
}
4.2 file-kafka配置文件
输入源为txt文件，输出源为kafka
input {
file {
codec => plain {
charset => "GB2312"
}
path => "D:/GameLog/BaseDir/*/*.txt"
discover_interval => 30
start_position => "beginning"
}
}
output {
kafka {
topic_id => "gamelog"
codec => plain {
format => "%{message}"
charset => "GB2312"
}
bootstrap_servers => "hadoop1:9092,hadoop2:9092,hadoop3:9092"
}
}
4.3 log-elasticsearch配置文件
输入源为nginx的日志文件，输出源为elasticsearch
input {
file {
type => "flow"
path => "var/logs/nginx/*.log"
discover_interval => 5
start_position => "beginning"
}
}
output {
if [type] == "flow" {
elasticsearch {
index => "flow-%{+YYYY.MM.dd}"
hosts => ["hadoop1:9200", "hadoop2:9200", "hadoop3:9200"]
}
}
}
4.4 kafka-elasticsearch配置文件
输入源为kafka的accesslog和gamelog主题，并在中间分别针对accesslog和gamelog进行过滤，输出源为elasticsearch。当input里面有多个kafka输入源时，client_id => "es*"必须添加且需要不同，否则会报错javax.management.InstanceAlreadyExistsException: kafka.consumer:type=app-info,id=logstash-0。
input {
kafka {
type => "accesslog"
codec => "plain"
auto_offset_reset => "earliest"
client_id => "es1"
group_id => "es1"
topics => ["accesslog"]
bootstrap_servers => "hadoop1:9092,hadoop2:9092,hadoop3:9092"
}
kafka {
type => "gamelog"
codec => "plain"
auto_offset_reset => "earliest"
client_id => "es2"
group_id => "es2"
topics => ["gamelog"]
bootstrap_servers => "hadoop1:9092,hadoop2:9092,hadoop3:9092"
}
}
filter {
if [type] == "accesslog" {
json {
source => "message"
remove_field => ["message"]
target => "access"
}
}
if [type] == "gamelog" {
mutate {
split => { "message" => " " }
add_field => {
"event_type" => "%{message[3]}"
"current_map" => "%{message[4]}"
"current_x" => "%{message[5]}"
"current_y" => "%{message[6]}"
"user" => "%{message[7]}"
"item" => "%{message[8]}"
"item_id" => "%{message[9]}"
"current_time" => "%{message[12]}"
}
remove_field => ["message"]
}
}
}
output {
if [type] == "accesslog" {
elasticsearch {
index => "accesslog"
codec => "json"
hosts => ["hadoop1:9200","hadoop2:9200","hadoop3:9200"]
}
}
if [type] == "gamelog" {
elasticsearch {
index => "gamelog"
codec => plain {
charset => "UTF-16BE"
}
hosts => ["hadoop1:9200","hadoop2:9200","hadoop3:9200"]
}
}
}
注：UTF-16BE为解决中文乱码，而不是UTF-8
5、logstash启动
logstash -f /opt/app/logstash-6.2.3/conf/flow-kafka.conf
6、logstash遇到的问题
1) 在使用logstash采集日志时，如果我们采用file为input类型，采用不能反复对一份文件进行测试！第一次会成功，之后就会失败！
参考资料：
https://blog.csdn.net/lvyuan1234/article/details/78653324