9

我有来自这种格式的日志。我已将 logstash 变量分配给下面的模式。我相信我已经使用随附的预定义 Grok 标签正确分配了这些元素中的每一个。但是，当我运行 logstash 时，它反映： [0] "_grokparsefailure" 表明它无法解析请求。我不知道我的conf到底出了什么问题。这里有人知道是什么原因造成的吗？我对logstash很陌生。提前致谢

1383834858 0 71.172.136.12 20097903 198.2.20.171 80 TCP_HIT/200 252 HEAD http://podcasts.someserver.com/80830A/podcasts.someserver.com/nyv/voice-film-club/2013/11/the-sexy-god -thor.mp3 - 0 355 "-" "播客/2.0" 33546 "-"

或者

%{BASE10NUM:timestamp} = 1383834858
%{BASE10NUM:time_taken} = 0
%{IP:clientip} = 71.172.136.12
%{BASE10NUM:filesize} = 20097903
%{IP:serverip} = 198.2.20.171
%{BASE10NUM:port} = 80
%{WORD:status_code} = TCP_HIT/200
%{BASE10NUM:sc_bytes} = 252
%{WORD:method} = HEAD
%{URI:cs_uri} = http://podcasts.someserver.com/80830A/podcasts.someserver.com/nyv/voice-   film-club/2013/11/the-sexy-god-thor.mp3
%{NOTSPACE:ignore2} = -
%{BASE10NUM:rs_duration} = 0
%{BASE10NUM:rs_bytes} = 355
%{QS:c_referrer} = "-"
%{QS:user_agent} = "Podcasts/2.0"
%{BASE10NUM:customerid} = 33546
%{QS:ignore} = "-"

我的 logstash.conf 文件如下所示：

input {
    #wpa_media logs from the CDN(see puppet module)
    redis {
        type => "wpc_media"
        host => "devredis1.somedomain.com"
        # these settings should match the output of the agent
        data_type => "list"
        key => "wpc_media"
        codec => json
        debug => true
   }
}


filter {
    grok {
        type    => "wpc_media"
        pattern => [ "%{BASE10NUM:timestamp} %{BASE10NUM:time_taken} %{IP:clientip} %{BASE10NUM:filesize} %{IP:serverip} %{BASE10NUM:port} %{WORD:status_code} %{BASE10NUM:sc_bytes} %{WORD:method} %{URI:cs_uri} %{NOTSPACE:ignore2} %{BASE10NUM:rs_duration} %{BASE10NUM:rs_bytes} %{QS:c_referrer} %{QS:user_agent} %{BASE10NUM:customerid} %{QS:ignore} " ]
    }

    mutate {
        #just something to cover up the error not really fixing it
        #remove_tag  => [ "_grokparsefailure" ]
        remove => [ "customer_id", "ignore", "c_referrer", "time_taken" ]
    }
}
output {
    stdout { debug => true debug_format => "ruby"}
}
4

1 回答 1

27

供您自己参考，GrokDebugger站点对于此类问题非常方便。

对于您提供的特定日志事件，%{WORD}不匹配TCP_HIT/200。

一种快速的解决方法是改为匹配%{DATA:status_code}（您可以在 GitHub 上查看内置模式）。你当然可以建立一个更有针对性的比赛，但如果没有看到可能的输入就很难做到。

如果您总是期待word/number，那么类似的东西(?<status_code>%{WORD}/%{INT})可能会起作用。

于 2013-11-07T22:54:20.757 回答