最近搭建一套日志监控平台，结合系统本身的特性总结一句话也就是：需要将Kafka中的数据导入到elasticsearch中。那么如何将Kafka中的数据导入到elasticsearch中去呢，总结起来大概有如下几种方式：
Kafka->logstash->elasticsearch->kibana(简单，只需启动一个代理程序)
Kafka->kafka-connect-elasticsearch->elasticsearch->kibana(与confluent绑定紧，有些复杂)
Kafka->elasticsearch-river-kafka-1.2.1-plugin->elasticsearch->kibana(代码很久没更新，后续支持比较差)
最终选择了第一种方案：
Logstash将topic数据写入es前可以做过滤ETL处理（总的来说还是比较好用的）
一、拓扑图
项目拓扑图如下所示：
此时消息的整体流向为：日志/消息整体流向Flume => kafka => logstash => elasticsearch => kibana（数据展示）
第一步：A .日志采集 Flume
Flume的指导工具书：
**Flume用户手册：============》**
https://flume.liyifeng.org/?spm=a2c4e.10696291.0.0.1c6e19a4oNUrUj#
agent.sources = r1
agent.channels = c1
agent.sinks = s1
agent.sources.r1.type = exec
agent.sources.r1.command = tail -F -n 0 /data01/monitorRequst.log
agent.sources.r1.restart = true //解决tail -F进程被杀死问题
agent.channels.c1.type = memory
agent.channels.c1.capacity = 1000
agent.channels.c1.transactionCapacity = 100
agent.sinks.s1.type = avro
agent.sinks.s1.port = 50001
agent.sinks.s1.hostname = IP
agent.sources.r1.channels = c1
agent.sinks.s1.channel = c1
Flume日志收集过程中踩过的坑可以参考：http://www.digitalsdp.com/Experiencebbs/maintenance/506.jhtml
B.Kafka Sink
agent.sources = r1
agent.channels = c2
agent.sinks = s2
agent.sources.r1.type = avro
agent.sources.r1.bind = IP
agent.sources.r1.port = 50001
agent.channels.c2.type = memory
agent.channels.c2.capacity = 1000
agent.channels.c2.transactionCapacity = 100
agent.sinks.s2.type = org.apace.flume.sink.kafka.KafkaSink
agent.sinks.s2.topic = XXX
agent.sinks.s2.brokerList = IP:9092,IP:9092
agent.sinks.s2.batchSize = 20
agent.sources.r1.channels = c2
agent.sinks.s2.channel = c2
1.7以后 Flume支持tailDir source
自己另外的一种配置：
a1.sources=r1
a1.channels=c1 c2
a1.sinks=k1 k2
# configure source
a1.sources.r1.type = TAILDIR
a1.sources.r1.positionFile = /root/module/flume/flume-data/taildir_positiontest.json
a1.sources.r1.filegroups = f1
a1.sources.r1.filegroups.f1 = /applog/gmall2019/log/gateway.log
a1.sources.r1.fileHeader = true
a1.sources.r1.channels = c1 c2
# configure channel
a1.channels.c1.type = memory
a1.channels.c1.capacity=10000
a1.channels.c1.byteCapacityBufferPercentage=20
a1.channels.ch1.type = file
a1.channels.c1.checkpointDir =/root/module/flume/flume-data/gla2020/app_ch1_tmp1
a1.channels.c1.dataDirs =/root/module/flume/flume-data/gla2020/app_ch1_tmpdata
a1.channels.c2.type = memory
a1.channels.c2.capacity=10000
a1.channels.c2.byteCapacityBufferPercentage=20
# configure sink
# gla_api-sink
a1.sinks.k1.type = org.apache.flume.sink.kafka.KafkaSink
a1.sinks.k1.kafka.topic = gateway_api
a1.sinks.k1.kafka.bootstrap.servers = node101:9092,node102:9092,node103:9092
a1.sinks.k1.kafka.flumeBatchSize = 2000
a1.sinks.k1.kafka.producer.acks = 1
a1.sinks.k1.channel = c1
a1.sources.r1.positionFile = /root/module/flume/flume-data/taildir_positiontest.json
*这里记录着数据采集到哪里，另外Flume是具有事务的，数据可做到不丢失，即使环境宕机，重新启动后依然可顺次进行数据采集工作。
*
二、环境搭建
关于Kafka及Flume的搭建在这里不再详细论述，如有需要请参见本文其它说明。在这里重点说明logstash的安装及配置。
A.下载logstash的安装包；
B.新建kafka-logstash-es.conf置于logstash/conf目录下；
C.配置kafka-logstash-es.conf如下：
logstash的配置语法如下：
input {
...#读取数据，logstash已提供非常多的插件，可以从file、redis、syslog等读取数据
}
filter{
...#想要从不规则的日志中提取关注的数据，就需要在这里处理。常用的有grok、mutate等
}
output{
...#输出数据，将上面处理后的数据输出到file、elasticsearch等
}
示例：
1：
input {
kafka {
zk_connect => "c1:2181,c2:2181,c3:2181"
group_id => "elasticconsumer" ---随意取
topic_id => "xxxlog" ---与flume中的Channel保持一致
reset_beginning => false
consumer_threads => 5
decorate_events => true
codec => "json"
}
}
output {
elasticsearch {
hosts => ["c4:9200","c5:9200"]
index => "traceid"--与Kafka中json字段无任何关联关系，注意：index必须小写
index => "log-%{+YYYY-MM-dd}"
workers => 5
codec => "json"
}
}
2:
input {
kafka {
bootstrap_servers => ["node101:9092"]
client_id => "kafka_esend"
group_id => "kafka_esend"
#auto_offset_reset => "latest"
auto_offset_reset => "earliest"
consumer_threads => 3
#decorate_events => true
topics => ["gla_api"]
#type => "bhy"
codec => "json"
}
}
filter {
}
output {
elasticsearch {
hosts => ["node101:9200"]
index => "gla_api"
#index => "log-%{+YYYY-MM-dd}"
#workers => 24
#codec => "json"
}
#stdout {
# JSON格式输出 控制台打印以json格式一行一行输出
# codec => json_lines
# }
}
示例2中，特别提示，如果不是为了测试看数据的写入情况，切记不要打印出数据，非常影响性能：
比如下面这个打印：（生产不要打印）
#stdout {
# JSON格式输出 控制台打印以json格式一行一行输出
# codec => json_lines
运行logstash命令为：nohup bin/logstash -f /XXX/logstash/conf/kafka-logstash-es.conf &
三、调测过程中遇到的一些坑
A.在集成ELK过程中总以为head插件是必须的，其实head插件为非必需品。elasticsearch仅提供了一个数据存储的煤介，head为了让大家更方便的去查看数据；
B.采用以上方案进行布署时，当系统正常运行时，可以在elasticsearch服务器上http://IP:9200/*中搜索index是否创建成功
参考：https://www.slahser.com/2016/04/21/日志监控平台搭建-关于Flume-Kafka-ELK/
http://www.jayveehe.com/2017/02/01/elk-stack/
http://wdxtub.com/2016/11/19/babel-log-analysis-platform-1/
**以下在举例几个Logstash的一些复杂配置============》**
供大家参考：
1 对数据进行过滤类型
conf1:
input {
kafka {
type => "ad"
bootstrap_servers => "114.118.13.66:9092,114.118.13.66:9093,114.118.13.66:9094"
client_id => "es_ad"
group_id => "es_ad"
auto_offset_reset => "latest" # 从最新的偏移量开始消费
consumer_threads => 5
decorate_events => true # 此属性会将当前topic、offset、group、partition等信息也带到message中
topics => ["ad_nginx_access_log"] # 数组类型，可配置多个topic
tags => ["nginx", "ad_access"]
}
}
filter {
if [type] == "ad" {
grok {
match => { "message" => "%{IPORHOST:clientip} - (%{USERNAME:user}|-) \[%{HTTPDATE:log_timestamp}\] \"%{WORD:http_method} %{NOTSPACE:request_uri} (?:HTTP\/\d\.\d)\" %{NUMBER:http_status} %{NUMBER:body_bytes_send} \"%{DATA:http_refer}\" \"%{DATA:user_agent}\" \"%{DATA:x_forword_for}\" %{NUMBER:action_length_time}" }
remove_field => ["message"]
}
date {
match => [ "log_timestamp" , "YYYY-MM-dd:HH:mm:ss Z" ]
}
geoip {
source => "clientip"
}
kv {
source => "request_uri"
field_split => "&?"
value_split => "="
include_keys => ["city_id","model","ad_position","osmodel","systemVersion","province_id","versionCode","mac","channel_number" ]
}
urldecode {
all_fields => true
}
}
}
output {
if [type] == "ad" {
elasticsearch {
hosts => ["114.118.10.253:9200"]
index => "adlog-%{+YYYY-MM-dd}"
document_type => "access_log"
timeout => 300
}
}
}
conf2：
input {
kafka {
codec => "plain"
group_id => "es2"
bootstrap_servers => ["kafka1:9092,kafka2:9092,kafka3:9092"] # 注意这里配置的kafka的broker地址不是zk的地址
auto_offset_reset => "earliest"
topics => ["gamekafka"]
}
}
filter {
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 {
elasticsearch {
index => "gamelogs"
codec => plain {
charset => "UTF-8"
}
hosts => ["elk1:9200", "elk2:9200", "elk3:9200"]
}
stdout {codec => rubydebug}
}
②启动写入到es的logstash
2 多个topic时Logstash的配置：
使用logstash-input-kafka消费信息并根据topic写入不同的es索引
分析：利用 decorate_events 参数，将 kafka的元信息（如：topic，group，offset等）全部输出，根据不同的topic即可输出到不同的es索引。
示例如下：
input{
kafka{
bootstrap_servers => ["55.0.10.19:9092,55.0.10.20:9092,55.0.10.21:9092"]
group_id => "kafkaToes5"
auto_offset_reset => "earliest"
consumer_threads => 5
decorate_events => true
topics => ["a1","a2"]
//security_protocol => "SASL_PLAINTEXT" #sasl认证使用
//sasl_mechanism => "PLAIN" #sasl认证使用
//jaas_path => "/var/vcap/jobs/logstash/bin/kafka.conf" #sasl认证使用-认证填写的密码文件
}
}
filter{
if [@metadata][kafka][topic] == "a1" {
json{
source => "message"
skip_on_invalid_json => true
}
mutate {
remove_field => ["@version","@timestamp"]
}
}
if [@metadata][kafka][topic] == "a2" {
json{
source => "message"
skip_on_invalid_json => true
}
mutate {
remove_field => ["@version","@timestamp"]
}
}
}
output {
stdout{
codec => json_lines
}
if [@metadata][kafka][topic] == "a1" {
elasticsearch {
index => "test123"
hosts => ["55.0.10.10:9280","55.0.10.11:9280"]
user => "admin"
password => "123456"
document_type=> "doc"
}
}
if [@metadata][kafka][topic] == "a2" {
elasticsearch {
index => "test456"
hosts => ["55.0.10.10:9280","55.0.10.11:9280"]
user => "admin"
password => "123456"
document_type=> "doc"
}
}
}
补充：
//认证填写的内容例子
// cat /var/vcap/jobs/logstash/bin/kafka.conf
KafkaClient {
org.apache.kafka.common.security.plain.PlainLoginModule required
username="kafkaUser"
password="kafkaPwd";
};
就先写到这里了 大家一起加油！！！
