[
https://issues.apache.org/jira/browse/SPARK-24987?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Yuval Itzchakov updated SPARK-24987:
------------------------------------
Description:
Setup:
* Spark 2.3.1
* Java 1.8.0 (112)
* Standalone Cluster Manager
* 3 Nodes, 1 Executor per node.
Spark 2.3.0 introduced a new mechanism for caching Kafka consumers
(https://issues.apache.org/jira/browse/SPARK-23623?page=com.atlassian.jira.plugin.system.issuetabpanels%3Aall-tabpanel)
via KafkaDataConsumer.acquire.
It seems that there are situations (I've been trying to debug it, haven't been
able to find the root cause as of yet) where cached consumers remain "in use"
throughout the life time of the task and are never released. This can be
identified by the following line of the stack trace:
at
org.apache.spark.sql.kafka010.KafkaDataConsumer$.acquire(KafkaDataConsumer.scala:460)
Which points to:
{code:java}
} else if (existingInternalConsumer.inUse) {
// If consumer is already cached but is currently in use, then return a new
consumer
NonCachedKafkaDataConsumer(newInternalConsumer)
{code}
Meaning the existing consumer created for that `TopicPartition` is still in
use for some reason. The weird thing is that you can see this for very old
tasks which have already finished successfully.
I've traced down this leak using file leak detector, attaching it to the
running Executor JVM process. I've emitted the list of open file descriptors
which [you can find
here|https://gist.github.com/YuvalItzchakov/cdbdd7f67604557fccfbcce673c49e5d],
and you can see that the majority of them are epoll FD used by Kafka Consumers,
indicating that they aren't closing.
Spark graph:
{code:java}
kafkaStream
.load()
.selectExpr("CAST(key AS STRING)", "CAST(value AS STRING)")
.as[(String, String)]
.flatMap {...}
.groupByKey(...)
.mapGroupsWithState(GroupStateTimeout.ProcessingTimeTimeout())(...)
.foreach(...)
.outputMode(OutputMode.Update)
.option("checkpointLocation",
sparkConfiguration.properties.checkpointDirectory)
.start()
.awaitTermination(){code}
was:
Spark 2.3.0 introduced a new mechanism for caching Kafka consumers
(https://issues.apache.org/jira/browse/SPARK-23623?page=com.atlassian.jira.plugin.system.issuetabpanels%3Aall-tabpanel)
via KafkaDataConsumer.acquire.
It seems that there are situations (I've been trying to debug it, haven't been
able to find the root cause as of yet) where cached consumers remain "in use"
throughout the life time of the task and are never released. This can be
identified by the following line of the stack trace:
at
org.apache.spark.sql.kafka010.KafkaDataConsumer$.acquire(KafkaDataConsumer.scala:460)
Which points to:
{code:java}
} else if (existingInternalConsumer.inUse) {
// If consumer is already cached but is currently in use, then return a new
consumer
NonCachedKafkaDataConsumer(newInternalConsumer)
{code}
Meaning the existing consumer created for that `TopicPartition` is still in
use for some reason. The weird thing is that you can see this for very old
tasks which have already finished successfully.
I've traced down this leak using file leak detector, attaching it to the
running Executor JVM process. I've emitted the list of open file descriptors
which [you can find
here|https://gist.github.com/YuvalItzchakov/cdbdd7f67604557fccfbcce673c49e5d],
and you can see that the majority of them are epoll FD used by Kafka Consumers,
indicating that they aren't closing.
Spark graph:
{code:java}
kafkaStream
.load()
.selectExpr("CAST(key AS STRING)", "CAST(value AS STRING)")
.as[(String, String)]
.flatMap {...}
.groupByKey(...)
.mapGroupsWithState(GroupStateTimeout.ProcessingTimeTimeout())(...)
.foreach(...)
.outputMode(OutputMode.Update)
.option("checkpointLocation",
sparkConfiguration.properties.checkpointDirectory)
.start()
.awaitTermination(){code}
> Kafka Cached Consumer Leaking Consumers
> ---------------------------------------
>
> Key: SPARK-24987
> URL: https://issues.apache.org/jira/browse/SPARK-24987
> Project: Spark
> Issue Type: Bug
> Components: Structured Streaming
> Affects Versions: 2.3.0, 2.3.1
> Environment: Spark 2.3.1
> Java(TM) SE Runtime Environment (build 1.8.0_112-b15)
> Java HotSpot(TM) 64-Bit Server VM (build 25.112-b15, mixed mode)
>
> Reporter: Yuval Itzchakov
> Priority: Critical
>
> Setup:
> * Spark 2.3.1
> * Java 1.8.0 (112)
> * Standalone Cluster Manager
> * 3 Nodes, 1 Executor per node.
> Spark 2.3.0 introduced a new mechanism for caching Kafka consumers
> (https://issues.apache.org/jira/browse/SPARK-23623?page=com.atlassian.jira.plugin.system.issuetabpanels%3Aall-tabpanel)
> via KafkaDataConsumer.acquire.
> It seems that there are situations (I've been trying to debug it, haven't
> been able to find the root cause as of yet) where cached consumers remain "in
> use" throughout the life time of the task and are never released. This can be
> identified by the following line of the stack trace:
> at
> org.apache.spark.sql.kafka010.KafkaDataConsumer$.acquire(KafkaDataConsumer.scala:460)
> Which points to:
> {code:java}
> } else if (existingInternalConsumer.inUse) {
> // If consumer is already cached but is currently in use, then return a new
> consumer
> NonCachedKafkaDataConsumer(newInternalConsumer)
> {code}
> Meaning the existing consumer created for that `TopicPartition` is still in
> use for some reason. The weird thing is that you can see this for very old
> tasks which have already finished successfully.
> I've traced down this leak using file leak detector, attaching it to the
> running Executor JVM process. I've emitted the list of open file descriptors
> which [you can find
> here|https://gist.github.com/YuvalItzchakov/cdbdd7f67604557fccfbcce673c49e5d],
> and you can see that the majority of them are epoll FD used by Kafka
> Consumers, indicating that they aren't closing.
> Spark graph:
> {code:java}
> kafkaStream
> .load()
> .selectExpr("CAST(key AS STRING)", "CAST(value AS STRING)")
> .as[(String, String)]
> .flatMap {...}
> .groupByKey(...)
> .mapGroupsWithState(GroupStateTimeout.ProcessingTimeTimeout())(...)
> .foreach(...)
> .outputMode(OutputMode.Update)
> .option("checkpointLocation",
> sparkConfiguration.properties.checkpointDirectory)
> .start()
> .awaitTermination(){code}
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
