[
https://issues.apache.org/jira/browse/FLINK-10774?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16687278#comment-16687278
]
ASF GitHub Bot commented on FLINK-10774:
----------------------------------------
stevenzwu edited a comment on issue #7020: [FLINK-10774] [Kafka] connection
leak when partition discovery is disabled an…
URL: https://github.com/apache/flink/pull/7020#issuecomment-438544127
we saw some exception regarding redundant calls to close method. will need
to fix it.
```
2018-11-13 18:26:54,928 ERROR
org.apache.flink.streaming.connectors.kafka.FlinkKafkaConsumerBase - failed to
close partitionDiscoverer
java.lang.IllegalStateException: This consumer has already been closed.
at
org.apache.kafka.clients.consumer.KafkaConsumer.ensureNotClosed(KafkaConsumer.java:1613)
at
org.apache.kafka.clients.consumer.KafkaConsumer.acquire(KafkaConsumer.java:1624)
at
org.apache.kafka.clients.consumer.KafkaConsumer.close(KafkaConsumer.java:1526)
at
org.apache.kafka.clients.consumer.KafkaConsumer.close(KafkaConsumer.java:1506)
at
org.apache.flink.streaming.connectors.kafka.internal.Kafka09PartitionDiscoverer.closeConnections(Kafka09PartitionDiscoverer.java:97)
at
org.apache.flink.streaming.connectors.kafka.internals.AbstractPartitionDiscoverer.close(AbstractPartitionDiscoverer.java:101)
at
org.apache.flink.streaming.connectors.kafka.FlinkKafkaConsumerBase.cancel(FlinkKafkaConsumerBase.java:673)
at
org.apache.flink.streaming.api.operators.StreamSource.cancel(StreamSource.java:108)
at
org.apache.flink.streaming.runtime.tasks.SourceStreamTask.cancelTask(SourceStreamTask.java:100)
at
org.apache.flink.streaming.runtime.tasks.StreamTask.cancel(StreamTask.java:369)
at
org.apache.flink.runtime.taskmanager.Task$TaskCanceler.run(Task.java:1481)
at java.lang.Thread.run(Thread.java:748)
```
Can `run` method and `cancel` method executed in different threads? if so,
we need to add `synchronized` to this method in `Kafka09PartitionDiscoverer`.
```
@Override
protected void closeConnections() throws Exception {
if (this.kafkaConsumer != null) {
this.kafkaConsumer.close();
// de-reference the consumer to avoid closing multiple
times
this.kafkaConsumer = null;
}
}
```
I looked at the `KafkaConsumer` code again. its close method calls
`acquire`. I thought it is acquire lock, which is not the case.
```
private void acquire() {
this.ensureNotClosed();
long threadId = Thread.currentThread().getId();
if (threadId != this.currentThread.get() &&
!this.currentThread.compareAndSet(-1L, threadId)) {
throw new ConcurrentModificationException("KafkaConsumer is not
safe for multi-threaded access");
} else {
this.refcount.incrementAndGet();
}
}
```
----------------------------------------------------------------
This is an automated message from the Apache Git Service.
To respond to the message, please log on GitHub and use the
URL above to go to the specific comment.
For queries about this service, please contact Infrastructure at:
us...@infra.apache.org
> connection leak when partition discovery is disabled and open throws exception
> ------------------------------------------------------------------------------
>
> Key: FLINK-10774
> URL: https://issues.apache.org/jira/browse/FLINK-10774
> Project: Flink
> Issue Type: Bug
> Components: Kafka Connector
> Affects Versions: 1.4.2, 1.5.5, 1.6.2
> Reporter: Steven Zhen Wu
> Assignee: Steven Zhen Wu
> Priority: Major
> Labels: pull-request-available
>
> Here is the scenario to reproduce the issue
> * partition discovery is disabled
> * open method throws an exception (e.g. when broker SSL authorization denies
> request)
> In this scenario, run method won't be executed. As a result,
> _partitionDiscoverer.close()_ won't be called. that caused the connection
> leak, because KafkaConsumer is initialized but not closed. That has caused
> outage that brought down our Kafka cluster, when a high-parallelism job got
> into a restart/failure loop.
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)
