With some WARN log, Yes, it is. and I found interesting things separately before I mentioned. I have another clusters. I run 2 brokers on 1 machine for test. and I see same problem before I mentioned, but I can’t see any error log on controller.log.
At this time, when I list topics with kafka-topic tool, I see information like below. == kafka-topics.sh =================================================== Topic:topicTRACE PartitionCount:2 ReplicationFactor:2 Configs:retention.ms=3600000 Topic: topicTRACE Partition: 0 Leader: 6 Replicas: 5,6 Isr: 6 Topic: topicTRACE Partition: 1 Leader: 6 Replicas: 6,5 Isr: 6,5 ====================================================================== but, producer keeps producing broker 5 which seems to be dead. when I get metadata from broker using my version of java api tool, info of leader and isr is different. Speaking properly, metadata from one broker is same as data from kafka-topics tool, metadata from another broker is different. ======================================================================== $ bin/kafka-run-class.sh com.kthcorp.daisy.ccprt.util.KafkaMetadata c-ccp-tk1-a60:9091 topicTRACE ---- topic info ---- partition: 0, leader: 5, replica: [id:5,host:c-ccp-tk1-a60,port:9091, id:6,host:c-ccp-tk1-a60,port:9092], isr: [id:5,host:c-ccp-tk1-a60,port:9091, id:6,host:c-ccp-tk1-a60,port:9092] partition: 1, leader: 6, replica: [id:6,host:c-ccp-tk1-a60,port:9092, id:5,host:c-ccp-tk1-a60,port:9091], isr: [id:6,host:c-ccp-tk1-a60,port:9092, id:5,host:c-ccp-tk1-a60,port:9091] ======================================================================== $ bin/kafka-run-class.sh com.kthcorp.daisy.ccprt.util.KafkaMetadata c-ccp-tk1-a60:9092 topicTRACE ---- topic info ---- partition: 0, leader: 6, replica: [id:5,host:c-ccp-tk1-a60,port:9091, id:6,host:c-ccp-tk1-a60,port:9092], isr: [id:6,host:c-ccp-tk1-a60,port:9092] partition: 1, leader: 6, replica: [id:6,host:c-ccp-tk1-a60,port:9092, id:5,host:c-ccp-tk1-a60,port:9091], isr: [id:6,host:c-ccp-tk1-a60,port:9092, id:5,host:c-ccp-tk1-a60,port:9091] ======================================================================== which one is correct? why is it happened? Thanks~ On Jun 10, 2014, at 11:28 PM, Jun Rao <jun...@gmail.com> wrote: > Ok. Was this host (broker id:1,host:c-ccp-tk1-a58,port:9091) up when the > controller had SocketTimeoutException? > > Thanks, > > Jun > > > On Mon, Jun 9, 2014 at 10:11 PM, Bongyeon Kim <bongyeon....@gmail.com> > wrote: > >> No, I can see any ZK session expiration log. >> >> What I have to do to prevent this? Increasing ' >> zookeeper.session.timeout.ms' >> can help? >> >> >> On Tue, Jun 10, 2014 at 12:58 PM, Jun Rao <jun...@gmail.com> wrote: >> >>> This is probably related to kafka-1382. The root cause is likely ZK >> session >>> expiration in the broker. Did you see any? >>> >>> Thanks, >>> >>> Jun >>> >>> >>> On Mon, Jun 9, 2014 at 8:11 PM, Bongyeon Kim <bongyeon....@gmail.com> >>> wrote: >>> >>>> Hi, team. >>>> >>>> I’m using 0.8.1. >>>> I found some strange log repeatedly on server.log in one of my brokers >>> and >>>> it keeps logging until now. >>>> >>>> server.log >>>> >>> >> ====================================================================================== >>>> ... >>>> [2014-06-09 10:41:47,402] ERROR Conditional update of path >>>> /brokers/topics/topicTRACE/partitions/1/state with data >>>> >>> >> {"controller_epoch":19,"leader":2,"version":1,"leader_epoch":43,"isr":[4,2]} >>>> and expected version 439 failed due to >>>> org.apache.zookeeper.KeeperException$BadVersionException: >>> KeeperErrorCode = >>>> BadVersion for /brokers/topics/topicTRACE/partitions/1/state >>>> (kafka.utils.ZkUtils$) >>>> [2014-06-09 10:41:47,402] INFO Partition [topicTRACE,1] on broker 2: >>>> Cached zkVersion [439] not equal to that in zookeeper, skip updating >> ISR >>>> (kafka.cluster.Partition) >>>> [2014-06-09 10:41:47,402] INFO Partition [topicDEBUG,0] on broker 2: >>>> Shrinking ISR for partition [topicDEBUG,0] from 1,3,2 to 2 >>>> (kafka.cluster.Partition) >>>> [2014-06-09 10:41:47,416] ERROR Conditional update of path >>>> /brokers/topics/topicDEBUG/partitions/0/state with data >>>> >>> >> {"controller_epoch":19,"leader":2,"version":1,"leader_epoch":43,"isr":[2]} >>>> and expected version 1424 failed due to >>>> org.apache.zookeeper.KeeperException$BadVersionException: >>> KeeperErrorCode = >>>> BadVersion for /brokers/topics/topicDEBUG/partitions/0/state >>>> (kafka.utils.ZkUtils$) >>>> [2014-06-09 10:41:47,432] INFO Partition [topicDEBUG,0] on broker 2: >>>> Cached zkVersion [1424] not equal to that in zookeeper, skip updating >> ISR >>>> (kafka.cluster.Partition) >>>> [2014-06-09 10:41:47,432] INFO Partition [topicCDR,3] on broker 2: >>>> Shrinking ISR for partition [topicCDR,3] from 4,1,2 to 2 >>>> (kafka.cluster.Partition) >>>> [2014-06-09 10:41:47,435] ERROR Conditional update of path >>>> /brokers/topics/topicCDR/partitions/3/state with data >>>> >>> >> {"controller_epoch":19,"leader":2,"version":1,"leader_epoch":46,"isr":[2]} >>>> and expected version 541 failed due to >>>> org.apache.zookeeper.KeeperException$BadVersionException: >>> KeeperErrorCode = >>>> BadVersion for /brokers/topics/topicCDR/partitions/3/state >>>> (kafka.utils.ZkUtils$) >>>> [2014-06-09 10:41:47,435] INFO Partition [topicCDR,3] on broker 2: >> Cached >>>> zkVersion [541] not equal to that in zookeeper, skip updating ISR >>>> (kafka.cluster.Partition) >>>> [2014-06-09 10:41:48,426] INFO Partition [topicTRACE,1] on broker 2: >>>> Shrinking ISR for partition [topicTRACE,1] from 4,3,2 to 4,2 >>>> (kafka.cluster.Partition) >>>> ... >>>> >>>> >>> >> ================================================================================================= >>>> >>>> and found some error and warning in controller.log >>>> >>>> >>>> controller.log >>>> >>> >> ====================================================================================== >>>> ... >>>> [2014-06-09 10:42:03,962] WARN [Controller-3-to-broker-1-send-thread], >>>> Controller 3 fails to send a request to broker >>>> id:1,host:c-ccp-tk1-a58,port:9091 (kafka.controller.RequestSendThread) >>>> java.net.SocketTimeoutException >>>> at >>>> sun.nio.ch.SocketAdaptor$SocketInputStream.read(SocketAdaptor.java:229) >>>> at >>> sun.nio.ch.ChannelInputStream.read(ChannelInputStream.java:103) >>>> at >>>> >>> >> java.nio.channels.Channels$ReadableByteChannelImpl.read(Channels.java:385) >>>> at kafka.utils.Utils$.read(Utils.scala:375) >>>> at >>>> >>> >> kafka.network.BoundedByteBufferReceive.readFrom(BoundedByteBufferReceive.scala:54) >>>> at >>>> kafka.network.Receive$class.readCompletely(Transmission.scala:56) >>>> at >>>> >>> >> kafka.network.BoundedByteBufferReceive.readCompletely(BoundedByteBufferReceive.scala:29) >>>> at >>> kafka.network.BlockingChannel.receive(BlockingChannel.scala:100) >>>> at >>>> >>> >> kafka.controller.RequestSendThread.doWork(ControllerChannelManager.scala:146) >>>> at >>> kafka.utils.ShutdownableThread.run(ShutdownableThread.scala:51) >>>> [2014-06-09 10:42:03,964] ERROR [Controller-3-to-broker-1-send-thread], >>>> Controller 3 epoch 21 failed to send UpdateMetadata request with >>>> correlation id 1 to broker id:1,host:c-ccp-tk1-a58,port:9091. >>> Reconnecting >>>> to broker. (kafka.controller.RequestSendThread) >>>> java.nio.channels.ClosedChannelException >>>> at kafka.network.BlockingChannel.send(BlockingChannel.scala:89) >>>> at >>>> >>> >> kafka.controller.RequestSendThread.liftedTree1$1(ControllerChannelManager.scala:132) >>>> at >>>> >>> >> kafka.controller.RequestSendThread.doWork(ControllerChannelManager.scala:131) >>>> at >>> kafka.utils.ShutdownableThread.run(ShutdownableThread.scala:51) >>>> >>>> ... >>>> >>>> [2014-06-09 10:42:38,064] WARN [OfflinePartitionLeaderSelector]: No >>> broker >>>> in ISR is alive for [topicTRACE,0]. Elect leader 3 from live brokers 3. >>>> There's potential data loss. >>>> (kafka.controller.OfflinePartitionLeaderSelector) >>>> ... >>>> >>>> >>> >> ================================================================================================= >>>> >>>> Why is this happen? Is there any possibilities data loss? >>>> To normalize my brokers, What I have to do? Do I have to restart this >>>> broker? >>>> >>>> >>>> Thanks in advance. >>>> >>>> >>>> >>> >> >> >> >> -- >> *Sincerely* >> *,**Bongyeon Kim* >> >> Java Developer & Engineer >> Seoul, Korea >> Mobile: +82-10-9369-1314 >> Email: bongyeon...@gmail.com >> Twitter: http://twitter.com/tigerby >> Facebook: http://facebook.com/tigerby >> Wiki: http://tigerby.com >>