elasticsearch - 使用logstash和 Elasticsearch 进行自定义解析

Question

我是Logtash的新手!
我配置好了，到目前为止一切正常。
我的日志文件如下:

2014-04-27 16:24:43 DEBUG b45e66 T+561   10.31.166.155   /v1/XXX<!session> XXX requested for category_ids: only_pro: XXX_ids:14525

如果我使用以下conf文件:

input { file { path => "/logs/*_log" }} output { elasticsearch { host => localhost } }

它将在ES中放置以下内容:

{
  _index: "logstash-2014.04.28",
  _type: "logs",
  _id: "WIoUbIvCQOqnz4tMZzMohg",
  _score: 1,
  _source: {
    message: "2014-04-27 16:24:43 DEBUG b45e66 T+561 10.31.166.155 This is my log !",
    @version: "1",
    @timestamp: "2014-04-28T14:25:52.165Z",
    host: "MYCOMPUTER",
    path: "\logs\xxx_app.log"
   }
  }

我如何在我的日志中获取字符串，以使整个文本不会在_source.message上？
例如:我希望我可以将其解析为:

 {
  _index: "logstash-2014.04.28",
  _type: "logs",
  _id: "WIoUbIvCQOqnz4tMZzMohg",
  _score: 1,
  _source: {
    logLevel: "DEBUG",
    messageId: "b45e66",
    sendFrom: "10.31.166.155",
    logTimestamp: "2014-04-27 16:24:43",
    message: "This is my log !",
    @version: "1",
    @timestamp: "2014-04-28T14:25:52.165Z",
    host: "MYCOMPUTER",
    path: "\logs\xxx_app.log"
   }
  }

Answer 1

您需要通过过滤器对其进行解析，例如grok filter。这可能有些棘手，所以请耐心尝试，尝试，尝试。并查看预定义的patterns。

您的消息的开始将是
%{DATESTAMP} %{WORD:logLevel} %{WORD:messageId} %{GREEDYDATA:someString} %{IP}
grokdebugger是一个非常有用的工具，可为您提供帮助。

完成后，您的配置应如下所示

input {
   stdin {}
}
filter {
  grok {
    match => { "message" => "%{DATESTAMP} %{WORD:logLevel} %{WORD:messageId} %{GREEDYDATA:someString} %{IP}" }
  }
}
output {
  elasticsearch { host => localhost }
}