You are viewing a plain text version of this content. The canonical link for it is here.
Posted to jira@kafka.apache.org by "David Mao (Jira)" <ji...@apache.org> on 2020/04/13 20:29:00 UTC
[jira] [Commented] (KAFKA-9189) Shutdown is blocked if connection
to Zookeeper is lost
[ https://issues.apache.org/jira/browse/KAFKA-9189?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17082642#comment-17082642 ]
David Mao commented on KAFKA-9189:
----------------------------------
also observed this issue, took a thread-dump and found that controlledShutdown() can block indefinitely in zkClient.getControllerId
{code:java}
"kafka-shutdown-hook" #24 prio=5 os_prio=0 cpu=9.96ms elapsed=184687.63s tid=0x00007f3b20786000 nid=0x4e69 waiting on condition [0x00007f3a8559a000]
java.lang.Thread.State: TIMED_WAITING (parking)
at jdk.internal.misc.Unsafe.park(java.base@11.0.5/Native Method)
- parking to wait for <0x00000006884f6808> (a java.util.concurrent.locks.AbstractQueuedSynchronizer$ConditionObject)
at java.util.concurrent.locks.LockSupport.parkNanos(java.base@11.0.5/LockSupport.java:234)
at java.util.concurrent.locks.AbstractQueuedSynchronizer$ConditionObject.awaitNanos(java.base@11.0.5/AbstractQueuedSynchronizer.java:2123)
at kafka.zookeeper.ZooKeeperClient.$anonfun$waitUntilConnected$3(ZooKeeperClient.scala:264)
at kafka.zookeeper.ZooKeeperClient.waitUntilConnected(ZooKeeperClient.scala:258)
at kafka.zookeeper.ZooKeeperClient.$anonfun$waitUntilConnected$1(ZooKeeperClient.scala:252)
at kafka.zookeeper.ZooKeeperClient.waitUntilConnected(ZooKeeperClient.scala:252)
at kafka.zk.KafkaZkClient.retryRequestsUntilConnected(KafkaZkClient.scala:1783)
at kafka.zk.KafkaZkClient.retryRequestsUntilConnected(KafkaZkClient.scala:1753)
at kafka.zk.KafkaZkClient.retryRequestUntilConnected(KafkaZkClient.scala:1748)
at kafka.zk.KafkaZkClient.getControllerId(KafkaZkClient.scala:1119)
at kafka.server.KafkaServer.doControlledShutdown$1(KafkaServer.scala:688)
at kafka.server.KafkaServer.controlledShutdown(KafkaServer.scala:772)
at kafka.server.KafkaServer.$anonfun$shutdown$2(KafkaServer.scala:794)
at kafka.server.KafkaServer$$Lambda$1583/0x0000000840924c40.apply$mcV$sp(Unknown Source)
at kafka.utils.CoreUtils$.swallow(CoreUtils.scala:68)
at kafka.server.KafkaServer.shutdown(KafkaServer.scala:794)
at kafka.server.KafkaServerStartable.shutdown(KafkaServerStartable.scala:54)
at kafka.Kafka$.$anonfun$main$3(Kafka.scala:80)
at kafka.Kafka$$$Lambda$172/0x00000008402a1440.apply$mcV$sp(Unknown Source)
at kafka.utils.Exit$.$anonfun$addShutdownHook$1(Exit.scala:38)
at kafka.Kafka$$$Lambda$173/0x00000008402a7840.run(Unknown Source)
at java.lang.Thread.run(java.base@11.0.5/Thread.java:834){code}
> Shutdown is blocked if connection to Zookeeper is lost
> ------------------------------------------------------
>
> Key: KAFKA-9189
> URL: https://issues.apache.org/jira/browse/KAFKA-9189
> Project: Kafka
> Issue Type: Bug
> Components: core
> Affects Versions: 2.3.0
> Environment: Linux, Docker 19.03.4
> Reporter: Boris Granveaud
> Priority: Minor
>
> We are using Kafka and Zookeeper in Docker swarm stacks. When we undeploy a stack, sometimes Kafka doesn't shutdown properly and is finally killed by Docker (thus leaving potentially corrupted files).
> Here are the steps to reproduce (simple Docker, no swarm):
>
> {code:java}
> docker network create test
> docker run -d --network test --name zk --rm zookeeper:3.5.6
> docker run --network test --name kf --rm -e "KAFKA_ZOOKEEPER_CONNECT=zk:2181" -e "KAFKA_ADVERTISED_LISTENERS=INSIDE://:9091" -e "KAFKA_LISTENER_SECURITY_PROTOCOL_MAP=INSIDE:PLAINTEXT" -e "KAFKA_INTER_BROKER_LISTENER_NAME=INSIDE" confluentinc/cp-kafka:5.3.1
> {code}
>
> In another shell:
>
> {code:java}
> docker stop zk
> docker stop kf
> {code}
>
--
This message was sent by Atlassian Jira
(v8.3.4#803005)