[
https://issues.apache.org/jira/browse/KAFKA-654?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13511579#comment-13511579
]
Joel Koshy commented on KAFKA-654:
----------------------------------
FWIW, it should be easy to reproduce this - I just saw it on a two node cluster
/ one topic but am looking at another issue right now.
[2012-12-06 10:10:18,380] ERROR [ReplicaFetcherThread-0-0-on-broker-1], Error
due to (kafka.server.ReplicaFetcherThread)
kafka.common.KafkaException: Trying to roll a new log segment for topic
partition abc-0 with start offset 0 while it already exsits
at kafka.log.Log.rollToOffset(Log.scala:456)
at kafka.log.Log.roll(Log.scala:434)
at kafka.log.Log.maybeRoll(Log.scala:423)
at kafka.log.Log.append(Log.scala:257)
at
kafka.server.ReplicaFetcherThread.processPartitionData(ReplicaFetcherThread.scala:51)
at
kafka.server.AbstractFetcherThread$$anonfun$doWork$5.apply(AbstractFetcherThread.scala:125)
at
kafka.server.AbstractFetcherThread$$anonfun$doWork$5.apply(AbstractFetcherThread.scala:108)
at scala.collection.immutable.Map$Map1.foreach(Map.scala:105)
at
kafka.server.AbstractFetcherThread.doWork(AbstractFetcherThread.scala:108)
at kafka.utils.ShutdownableThread.run(ShutdownableThread.scala:50)
[2012-12-06 10:10:18,382] INFO [ReplicaFetcherThread-0-0-on-broker-1], Stopped
(kafka.server.ReplicaFetcherThread)
> Irrecoverable error while trying to roll a segment that already exists
> ----------------------------------------------------------------------
>
> Key: KAFKA-654
> URL: https://issues.apache.org/jira/browse/KAFKA-654
> Project: Kafka
> Issue Type: Bug
> Affects Versions: 0.8
> Reporter: Neha Narkhede
> Priority: Blocker
>
> I tried setting up a 5 broker 0.8 cluster and sending messages to 100s of
> topics on it. For a couple of topic partitions, the produce requests never
> succeed since they fail on the leader with the following error -
> [2012-12-05 22:54:05,711] WARN [Kafka Log on Broker 2], Newly rolled segment
> file 0000000000000000000
> 0.log already exists; deleting it first (kafka.log.Log)
> [2012-12-05 22:54:05,711] WARN [Kafka Log on Broker 2], Newly rolled segment
> file 0000000000000000000
> 0.index already exists; deleting it first (kafka.log.Log)
> [2012-12-05 22:54:05,715] ERROR [ReplicaFetcherThread-1-0-on-broker-2], Error
> due to (kafka.server.R
> eplicaFetcherThread)
> kafka.common.KafkaException: Trying to roll a new log segment for topic
> partition NusWriteEvent-4 with start offset 0 while it already exsits
> at kafka.log.Log.rollToOffset(Log.scala:456)
> at kafka.log.Log.roll(Log.scala:434)
> at kafka.log.Log.maybeRoll(Log.scala:423)
> at kafka.log.Log.append(Log.scala:257)
> at
> kafka.server.ReplicaFetcherThread.processPartitionData(ReplicaFetcherThread.scala:51)
> at
> kafka.server.AbstractFetcherThread$$anonfun$doWork$5.apply(AbstractFetcherThread.scala:125)
> at
> kafka.server.AbstractFetcherThread$$anonfun$doWork$5.apply(AbstractFetcherThread.scala:108)
> at
> scala.collection.immutable.HashMap$HashMap1.foreach(HashMap.scala:125)
> at
> scala.collection.immutable.HashMap$HashTrieMap.foreach(HashMap.scala:344)
> at
> scala.collection.immutable.HashMap$HashTrieMap.foreach(HashMap.scala:344)
> at
> kafka.server.AbstractFetcherThread.doWork(AbstractFetcherThread.scala:108)
> at kafka.utils.ShutdownableThread.run(ShutdownableThread.scala:50)
--
This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators
For more information on JIRA, see: http://www.atlassian.com/software/jira
