[KAFKA-6252] A metric named 'XX' already exists, can't register another one.

• Type: Bug
• Status: Resolved
• Priority: Critical
• Resolution: Fixed
• Affects Version/s: 1.0.0
• Fix Version/s: 1.0.1, 1.1.0
• Component/s: connect
• Environment: Linux
• Assignee: Arjun Satish
• Reporter: Alexis Sellier
• Reviewer: Ewen Cheslack-Postava
• Created: 21/Nov/17 17:03
• Updated: 01/Feb/18 10:27
• Resolved: 06/Jan/18 05:25

Description

When a connector crashes (or is not implemented correctly by not stopping/interrupting poll()), It cannot be restarted and an exception like this is thrown

java.lang.IllegalArgumentException: A metric named 'MetricName [name=offset-commit-max-time-ms, group=connector-task-metrics, description=The maximum time in milliseconds taken by this task to commit offsets., tags={connector=hdfs-sink-connector-recover, task=0}]' already exists, can't register another one.
	at org.apache.kafka.common.metrics.Metrics.registerMetric(Metrics.java:532)
	at org.apache.kafka.common.metrics.Sensor.add(Sensor.java:256)
	at org.apache.kafka.common.metrics.Sensor.add(Sensor.java:241)
	at org.apache.kafka.connect.runtime.WorkerTask$TaskMetricsGroup.<init>(WorkerTask.java:328)
	at org.apache.kafka.connect.runtime.WorkerTask.<init>(WorkerTask.java:69)
	at org.apache.kafka.connect.runtime.WorkerSinkTask.<init>(WorkerSinkTask.java:98)
	at org.apache.kafka.connect.runtime.Worker.buildWorkerTask(Worker.java:449)
	at org.apache.kafka.connect.runtime.Worker.startTask(Worker.java:404)
	at org.apache.kafka.connect.runtime.distributed.DistributedHerder.startTask(DistributedHerder.java:852)
	at org.apache.kafka.connect.runtime.distributed.DistributedHerder.access$1600(DistributedHerder.java:108)
	at org.apache.kafka.connect.runtime.distributed.DistributedHerder$13.call(DistributedHerder.java:866)
	at org.apache.kafka.connect.runtime.distributed.DistributedHerder$13.call(DistributedHerder.java:862)
	at java.util.concurrent.FutureTask.run(FutureTask.java:266)
	at java.util.concurrent.ThreadPoolExecutor.runWorker(ThreadPoolExecutor.java:1142)
	at java.util.concurrent.ThreadPoolExecutor$Worker.run(ThreadPoolExecutor.java:617)
	at java.lang.Thread.run(Thread.java:745)

I guess it's because the function taskMetricsGroup.close is not call in all the cases

Issue Links

• is duplicated by: KAFKA-6279 Connect metrics do not get cleaned up for a source connector that doesn't stop properly
• is duplicated by: KAFKA-6466 Kafka connect task sometimes fails on start-up
• links to: GitHub Pull Request #4397