Filebeat、Logstash、Kafka 三步曲

Filebeat > Logstash > Kafka 使用步骤

使用filebeat收集日志到logstash中,再由logstash再生产数据到kafka,如果kafka那边没有kerberos认证也可以直接收集到kafka中。

使用方法

PS:截至目前时间2018-09-02为止logstash的版本为6.4.0有Integer转Long的Bug,官方说预计会在本月修复，所以这里先降低一下logstash的版本,暂时使用6.2.2。

下载logstash

cd /usr/share
wget https://artifacts.elastic.co/downloads/logstash/logstash-6.2.2.tar.gz
mv logstash-6.2.2 logstash

Filebeat

安装filebeat

yum install filebeat

filebeat配置文件 /etc/filebeat/filebeat.yml

filebeat.inputs:
- type: log
  enabled: true
  publish_async: true
  paths:    ## 日志位置
    - /etc/filebeat/logs/*.logfilebeat.config.modules:
  path: ${path.config}/modules.d/*.yml
  reload.enabled: falsesetup.template.settings:
  index.number_of_shards: 3
output.logstash:
  hosts: ["127.0.0.1:5044"]
processors:
 - drop_fields:
     fields: []

---

Logstash

安装kafka插件

/usr/share/logstash/bin/logstash-plugin install logstash-output-kafka

配置logstash

mkdir /etc/logstash

logstash pipelines文件 /etc/logstash/pipelines.yml

 - pipeline.id: another_test   queue.type: persisted
   path.config: "/etc/logstash/conf.d/*.conf"
   pipeline.id: mylogs
   pipeline.workers: 8
   pipeline.batch.size: 1000
   pipeline.batch.delay: 50
   pipeline.output.workers: 1

logstash 配置文件 /etc/logstash/logstash.yml

   pipeline:
     batch:
       size: 125
       delay: 5
       path.data: /var/lib/logstash
pipeline.workers: 8pipeline.batch.size: 1000path.logs: /var/log/logstash

jvm 配置 /etc/logstash/jvm.options

-Xms10g
-Xmx10g
-XX:+UseParNewGC
-XX:+UseConcMarkSweepGC
-XX:CMSInitiatingOccupancyFraction=75-XX:+UseCMSInitiatingOccupancyOnly
-Djava.awt.headless=true-Dfile.encoding=UTF-8-Djruby.compile.invokedynamic=true-Djruby.jit.threshold=0-XX:+HeapDumpOnOutOfMemoryError
-Djava.security.egd=file:/dev/urandom

/etc/logstash/conf.d/kafka/kafka_jaas.conf

KafkaClient {
  com.sun.security.auth.module.Krb5LoginModule required
  doNotPrompt=true
  useTicketCache=true
  principal="admin/[e-mail]"
  useKeyTab=true
  serviceName="kafka"
  keyTab="/etc/security/keytabs/admin.keytab"
  client=true;
};

kafka认证文件 /etc/logstash/conf.d/logstash.conf

input {
    beats{
        port => 5044
    }
}
output{
    kafka {
        topic_id => "test_hello"
        bootstrap_servers => "t2.demo.com:9092,t3.demo.com:9092,t4.demo.com:9092"
        compression_type => "snappy"
        jaas_path => "/etc/logstash/conf.d/kafka/kafka_jaas.conf"
        kerberos_config => "/etc/krb5.conf"
        ##名字一定要跟kafka_jaas.conf里面的serviceName一样
        sasl_kerberos_service_name => "kafka"
        security_protocol => "SASL_PLAINTEXT"
        client_id => "log.demo.com"
    }
}

运行服务

运行logstash

/usr/share/logstash/bin/logstash --path.settings /etc/logstash

运行filebeat

systemctl start filebeat

日志

logstash 正常运行日志

[root@log logstash]# /usr/share/logstash/bin/logstash --path.settings /etc/logstash
Could not find log4j2 configuration at path /etc/logstash/log4j2.properties. Using default config which logs errors to the console
[INFO ] 2018-09-02 23:33:27.758 [main] scaffold - Initializing module {:module_name=>"netflow", :directory=>"/usr/share/logstash/modules/netflow/configuration"}
[INFO ] 2018-09-02 23:33:27.772 [main] scaffold - Initializing module {:module_name=>"fb_apache", :directory=>"/usr/share/logstash/modules/fb_apache/configuration"}
[INFO ] 2018-09-02 23:33:28.657 [LogStash::Runner] runner - Starting Logstash {"logstash.version"=>"6.2.2"}
[INFO ] 2018-09-02 23:33:29.170 [Api Webserver] agent - Successfully started Logstash API endpoint {:port=>9600}
[INFO ] 2018-09-02 23:33:31.075 [Ruby-0-Thread-1: /usr/share/logstash/lib/bootstrap/environment.rb:6] pipeline - Starting pipeline {:pipeline_id=>"mylogs", "pipeline.workers"=>8, "pipeline.batch.size"=>1000, "pipeline.batch.delay"=>50}
[INFO ] 2018-09-02 23:33:31.199 [[mylogs]-pipeline-manager] ProducerConfig - ProducerConfig values: 
    acks = 1
    batch.size = 16384
    bootstrap.servers = [t2.demo.com:9092, t3.demo.com:9092, t4.demo.com:9092]
    buffer.memory = 33554432
    client.id = log.demo.com
    compression.type = snappy
    connections.max.idle.ms = 540000
    enable.idempotence = false
    interceptor.classes = null
    key.serializer = class org.apache.kafka.common.serialization.StringSerializer
    linger.ms = 0
    max.block.ms = 60000
    max.in.flight.requests.per.connection = 5
    max.request.size = 1048576
    metadata.max.age.ms = 300000
    metric.reporters = []
    metrics.num.samples = 2
    metrics.recording.level = INFO
    metrics.sample.window.ms = 30000
    partitioner.class = class org.apache.kafka.clients.producer.internals.DefaultPartitioner
    receive.buffer.bytes = 32768
    reconnect.backoff.max.ms = 10
    reconnect.backoff.ms = 10
    request.timeout.ms = 30000
    retries = 0
    retry.backoff.ms = 100
    sasl.jaas.config = null
    sasl.kerberos.kinit.cmd = /usr/bin/kinit
    sasl.kerberos.min.time.before.relogin = 60000
    sasl.kerberos.service.name = kafka
    sasl.kerberos.ticket.renew.jitter = 0.05
    sasl.kerberos.ticket.renew.window.factor = 0.8
    sasl.mechanism = GSSAPI
    security.protocol = SASL_PLAINTEXT
    send.buffer.bytes = 131072
    ssl.cipher.suites = null
    ssl.enabled.protocols = [TLSv1.2, TLSv1.1, TLSv1]
    ssl.endpoint.identification.algorithm = null
    ssl.key.password = null
    ssl.keymanager.algorithm = SunX509
    ssl.keystore.location = null
    ssl.keystore.password = null
    ssl.keystore.type = JKS
    ssl.protocol = TLS
    ssl.provider = null
    ssl.secure.random.implementation = null
    ssl.trustmanager.algorithm = PKIX
    ssl.truststore.location = null
    ssl.truststore.password = null
    ssl.truststore.type = JKS
    transaction.timeout.ms = 60000
    transactional.id = null
    value.serializer = class org.apache.kafka.common.serialization.StringSerializer

[INFO ] 2018-09-02 23:33:31.311 [[mylogs]-pipeline-manager] AbstractLogin - Successfully logged in.
[INFO ] 2018-09-02 23:33:31.315 [kafka-kerberos-refresh-thread-admin/[e-mail]] KerberosLogin - [Principal=admin/[e-mail]]: TGT refresh thread started.
[INFO ] 2018-09-02 23:33:31.317 [kafka-kerberos-refresh-thread-admin/[e-mail]] KerberosLogin - [Principal=admin/[e-mail]]: TGT valid starting at: 2018-09-02T23:33:31.000+0800[INFO ] 2018-09-02 23:33:31.317 [kafka-kerberos-refresh-thread-admin/[e-mail]] KerberosLogin - [Principal=admin/[e-mail]]: TGT expires: 2018-09-03T23:33:31.000+0800[INFO ] 2018-09-02 23:33:31.317 [kafka-kerberos-refresh-thread-admin/[e-mail]] KerberosLogin - [Principal=admin/[e-mail]]: TGT refresh sleeping until: 2018-09-03T19:41:17.691+0800[INFO ] 2018-09-02 23:33:31.338 [[mylogs]-pipeline-manager] AppInfoParser - Kafka version : 1.0.0[INFO ] 2018-09-02 23:33:31.338 [[mylogs]-pipeline-manager] AppInfoParser - Kafka commitId : aaa7af6d4a11b29d
[INFO ] 2018-09-02 23:33:31.653 [[mylogs]-pipeline-manager] beats - Beats inputs: Starting input listener {:address=>"0.0.0.0:5044"}
[INFO ] 2018-09-02 23:33:31.705 [Ruby-0-Thread-1: /usr/share/logstash/lib/bootstrap/environment.rb:6] pipeline - Pipeline started succesfully {:pipeline_id=>"mylogs", :thread=>"#<Thread:0x49c5d932 run>"}
[INFO ] 2018-09-02 23:33:31.752 [[mylogs]<beats] Server - Starting server on port: 5044[INFO ] 2018-09-02 23:33:31.780 [Ruby-0-Thread-1: /usr/share/logstash/lib/bootstrap/environment.rb

作者：dounine
链接：https://www.jianshu.com/p/97de40a5e2dc