基于日志消息查询日志的日志分析

Question

我有一个Java应用程序，它以这种格式输出日志

时间戳UUID1一些信息

时间戳UUID1更多信息

时间戳UUID1 x=1

时间戳UUID2一些信息

时间戳UUID2更多信息

时间戳UUID2 x=2

时间戳UUID3一些信息

时间戳UUID3更多信息

时间戳UUID3 x=1

我想用Elsatic、LogStash和Kibana实现一个日志分析框架。是否有可能只根据X值获取日志？

例如：-

如果我查询X= 1，我应该只获得以下日志。

时间戳UUID1一些信息

时间戳UUID1更多信息

时间戳UUID1 x=1

时间戳UUID3一些信息

时间戳UUID3更多信息

时间戳UUID3 x=1

如果我查询X= 2，我应该只获得以下日志。

时间戳UUID2一些信息

时间戳UUID2更多信息

时间戳UUID2 x=2

我控制日志消息格式。如果不能直接执行此查询，我也可以更改消息格式。

更新1：

我会说得更具体一点。

下面是我的日志语句。

    MDC.put("uuid", UUID.randomUUID().toString());
    logger.info("Assigning value to the variable : {}", name);
    this.setVal(value.getVal());
    logger.info("{} = {}", name, value.getVal());
    logger.info("Assigned value {} to the variable : {}", value.getVal(),
            name);
    MDC.clear();

我使用UDP在Logstash中接收到日志语句。我收到的信息就像。

​

{
     "@timestamp" => "2015-04-01T10:23:37.846+05:30",
       "@version" => 1,
        "message" => "Assigning value to the variable : X",
    "logger_name" => "com.example.logstash.Variable",
    "thread_name" => "pool-1-thread-1",
          "level" => "INFO",
    "level_value" => 20000,
       "HOSTNAME" => "pnibinkj-W7-1",
           "uuid" => "ab17b842-8348-4474-98e4-8bc2b8dd6781",
           "host" => "127.0.0.1"
}
{
     "@timestamp" => "2015-04-01T10:23:37.846+05:30",
       "@version" => 1,
        "message" => "Assigning value to the variable : Y",
    "logger_name" => "com.example.logstash.Variable",
    "thread_name" => "pool-1-thread-2",
          "level" => "INFO",
    "level_value" => 20000,
       "HOSTNAME" => "pnibinkj-W7-1",
           "uuid" => "d5513e4c-de3b-4144-87e4-87b077ac8056",
           "host" => "127.0.0.1"
}
{
     "@timestamp" => "2015-04-01T10:23:37.862+05:30",
       "@version" => 1,
        "message" => "Y = 1",
    "logger_name" => "com.example.logstash.Variable",
    "thread_name" => "pool-1-thread-2",
          "level" => "INFO",
    "level_value" => 20000,
       "HOSTNAME" => "pnibinkj-W7-1",
           "uuid" => "d5513e4c-de3b-4144-87e4-87b077ac8056",
           "host" => "127.0.0.1"
}
{
     "@timestamp" => "2015-04-01T10:23:37.863+05:30",
       "@version" => 1,
        "message" => "X = 1",
    "logger_name" => "com.example.logstash.Variable",
    "thread_name" => "pool-1-thread-1",
          "level" => "INFO",
    "level_value" => 20000,
       "HOSTNAME" => "pnibinkj-W7-1",
           "uuid" => "ab17b842-8348-4474-98e4-8bc2b8dd6781",
           "host" => "127.0.0.1"
}
{
     "@timestamp" => "2015-04-01T10:23:37.863+05:30",
       "@version" => 1,
        "message" => "Assigned value 1 to the variable : X",
    "logger_name" => "com.example.logstash.Variable",
    "thread_name" => "pool-1-thread-1",
          "level" => "INFO",
    "level_value" => 20000,
       "HOSTNAME" => "pnibinkj-W7-1",
           "uuid" => "ab17b842-8348-4474-98e4-8bc2b8dd6781",
           "host" => "127.0.0.1"
}
{
     "@timestamp" => "2015-04-01T10:23:37.863+05:30",
       "@version" => 1,
        "message" => "Assigned value 1 to the variable : Y",
    "logger_name" => "com.example.logstash.Variable",
    "thread_name" => "pool-1-thread-2",
          "level" => "INFO",
    "level_value" => 20000,
       "HOSTNAME" => "pnibinkj-W7-1",
           "uuid" => "d5513e4c-de3b-4144-87e4-87b077ac8056",
           "host" => "127.0.0.1"
}

​

有两个UUID

"d5513e4c-de3b-4144-87e4-87b077ac8056“，用于"Y = 1”

"ab17b842-8348-4474-98e4-8bc2b8dd6781“(X=1)

每个UUID还有另外两条消息。我想把它们合并成一个单一的事件。

我不确定，如何为这种情况编写多行过滤器。

​

filter {
  multiline {
    pattern => "."
    what => "previous"
    stream_identity => "%{uuid}"
  }
}

​

“模式”和“什么”似乎是必需的字段。我应该为这些领域准备些什么。如何使用流标识？

请给我指出正确的方向。

谢谢，保罗

Answer 1

您需要组合您的消息(请参阅支持stream_identity的多行{}过滤器)，然后常规查询将返回适当的消息。

Answer 2

如果X是一些唯一的值，这应该可以使用kibana过滤器，但是对于显示格式的日志，您需要使用多行筛选器将条目连接在一起。

有了这一点，您可能会使用类似于

message: "X=1"