Kafka input plugin creates anywhere from 2-5x more documents than in topic

Elastic Stack Logstash
1

I am trying to run a small pool of logstash with at-most-once delivery semantics on a kafka 0.10.2.1 cluster. The kafka input client is running with the new style of kafka consumer group, managed as metadata in the brokers, and it does not use Zookeeper for offset tracking. I have a brief commit interval to keep the offsets moving between consumers.

input {
  kafka {
    bootstrap_servers => "${KAFKA_BOOTSTRAP_SERVERS}"
    topics => ["proofpoint", "fortigate", "windows"]
    group_id => "logstash"
    auto_offset_reset => "latest"
    auto_commit_interval_ms => "50"

    codec => "json"
    consumer_threads => 1
    partition_assignment_strategy => "org.apache.kafka.clients.consumer.RoundRobinAssignor"
  }
}

My elasticsearch output config:

output {
    elasticsearch {
      hosts => ["${ELASTICSEARCH_CLIENT_HOST}"]
      index => "@%{vtype}-nad-v2-%{date}"
      flush_size => 1000
      idle_flush_time => 10
      sniffing => true
      sniffing_delay => 30
    }
}

Now I see in Elasticsearch that I have 5x the data from what is in kafka. What am I doing wrong?

2

This topic was automatically closed 28 days after the last reply. New replies are no longer allowed.