Would someone please paste a verbose sample of filter metrics?

Elastic Stack Logstash
1

I'm analyzing nginx logs with ELK.

For the logs were too large (10GB per day), I don't want to save all the detail data into es, but only the aggregations.And I found metrics filter is suitable for my occasion.

What I wanna record was the counts of different groups for every ten minutes.

  • counts of groups (ip, user-agent)
  • counts of groups (ip, session_id, platform)
  • and so on

However, the document of metric filter is too simple for me to understand.
This is the config file I've written, it seemed not work.

input {
    file {
        path => "/path/to/access.log"
    }
}

filter {
    some_filter { 
        # grok,geoip,useragent to parse log
    }
    metrics { 
        add_field => { 
            "session_id" => "%{id}" 
            "time" => "%{t}" 
            "landing_page" => "%{lh}"
            "period" => "%{p}" 
            "platform" => "%{pf}" 
        } 
        timer => ["aggs", "%{id} %{t} %{lh} %{p} %{pf}"]
        add_tag => "me"
    }
}

output {
    if "me" in [tags] {
    elasticsearch {
        index => "user_record"
        document_type => "tjj_access"
        hosts => ["10.10.20.143:9200"]
    }
    #stdout { codec => json }
    }
}