Skip to content

zigarn/kafka-connect-jmx

Repository files navigation

JMX Source Connector

The JmxSourceConnector is a Source Connector that collects JMX metrics and push them to Kafka.

Build status: CircleCI

Configuration

Name Type Importance Default Value Validator Documentation
topic String High The topic to publish data to
jmx.url String High The JMX URL to fetch data from
jmx.username String Medium "" The username to connect to JMX
jmx.password String Medium [hidden] The password to connect to JMX
connection.attempts Int Low 3 [0,...] Maximum number of attempts to retrieve a JMX connection
connection.backoff.ms Long Low 10 000 [0,...] Backoff time in milliseconds between connection attempts

Connect-standalone example

This configuration is used typically along with standalone mode:

name=jmx-connector
tasks.max=1

connector.class=com.zigarn.kafka.connect.jmx.JmxSourceConnector
topic=connect-jmx
jmx.url=service:jmx:rmi:///jndi/rmi://localhost:9999/jmxrmi
# Set the following configuration if JMX authentication is necessary
jmx.username=jmx_user
jmx.password=jmx_password

Connect-distributed

This configuration is used typically along with distributed mode:

{
    "name": "jmx-connector",
    "config": {
        "connector.class": "com.zigarn.kafka.connect.jmx.JmxSourceConnector",
        "topic": "connect-jmx",
        "jmx.url": "service:jmx:rmi:///jndi/rmi://localhost:9999/jmxrmi",
        "connection.attempts": 5,
        "connection.backoff.ms": 1000
    }
}

Put this configuration in a connector.json file, then post it to one of the Kafka-connect workers:

curl -s -X POST -H 'Content-Type: application/json' --data @connector.json http://localhost:8083/connectors

Output

For each MBean available on the JMX server, a record will be sent to the configured topic with the following structure:

  • key: Struct with fields:
    • jmx.url: JMX service URL the MBean was retrieved from
    • bean: canonical name of the MBean
    • timestamp: timestamp when the MBean was retrieved
  • value: a map of MBean attributes with attribute name as key and String representation of attribute value as value

For example:

// Key
{
  "jmx.url":"/jndi/rmi://localhost:9999/jmxrmi",
  "bean":"kafka.producer:client-id=confluent-control-center-heartbeat-sender-1-producer,type=producer-metrics",
  "timestamp":1516713540685
}
// Value
{
  "connection-creation-total":"4.0",
  "bufferpool-wait-time-total":"0.0",
  "batch-split-total":"0.0",
  "produce-throttle-time-max":"0.0",
  "select-rate":"0.3663979748184665",
  "connection-close-total":"0.0",
  "outgoing-byte-rate":"113.48899714429699",
  "record-send-total":"300.0",
  "batch-size-max":"137.0",
  "produce-throttle-time-avg":"0.0",
  "iotime-total":"6.6670434E7",
  "successful-authentication-total":"4.0",
  "batch-split-rate":"0.0",
  "io-waittime-total":"2.5733506471E11",
  "request-rate":"0.06719075455217363",
  "buffer-available-bytes":"3.3554432E7",
  ...
  "io-time-ns-avg":"90059.09090909091",
  "compression-rate-avg":"0.9496202588081359",
  "record-retry-rate":"0.0",
  "request-latency-max":"104.0",
  "record-size-max":"157.0",
  "select-total":"2.5733506471E11",
  "buffer-total-bytes":"3.3554432E7",
  "batch-size-avg":"136.4"
}

Further documentation

Please see documentation.

License

Apache License, Version 2.0