Dan Swanson created KAFKA-975:
---------------------------------
Summary: Leader not local for partition when partition is leader
Key: KAFKA-975
URL: https://issues.apache.org/jira/browse/KAFKA-975
Project: Kafka
Issue Type: Bug
Components: replication
Affects Versions: 0.8
Environment: centos 6.4
Reporter: Dan Swanson
Assignee: Neha Narkhede
I have a two server kafka cluster (dev003 and dev004). I am following the
example from this URL but using two servers with a single kafka instance
instead of using 1 server with two instances..
http://www.michael-noll.com/blog/2013/03/13/running-a-multi-broker-apache-kafka-cluster-on-a-single-node/
Using the following trunk version
commit c27c768463a5dc6be113f2e5b3e00bf8d9d9d602
Author: David Arthur <mum...@gmail.com>
Date: Thu Jul 11 15:34:57 2013 -0700
KAFKA-852, remove clientId from Offset{Fetch,Commit}Response. Reviewed by
Jay.
------
[2013-07-16 10:56:50,279] INFO [Kafka Server 3], started
(kafka.server.KafkaServer)
------
dan@linux-rr29:~/git-data/kafka-current-src> bin/kafka-topics.sh --zookeeper
dev003:2181 --create --topic dadj1 --partitions 1 --replication-factor 2
2>/dev/null
Created topic "dadj1".
dan@linux-rr29:~/git-data/kafka-current-src>
-------
[2013-07-16 10:56:57,946] INFO [Replica Manager on Broker 3]: Handling
LeaderAndIsr request
Name:LeaderAndIsrRequest;Version:0;Controller:4;ControllerEpoch:19;CorrelationId:12;ClientId:id_4-host_dev004-port_9092;PartitionState:(dadj1,0)
->
(LeaderAndIsrInfo:(Leader:3,ISR:3,4,LeaderEpoch:0,ControllerEpoch:19),ReplicationFactor:2),AllReplicas:3,4);Leaders:id:3,host:dev003,port:9092
(kafka.server.ReplicaManager)
[2013-07-16 10:56:57,959] INFO [ReplicaFetcherManager on broker 3] Removing
fetcher for partition [dadj1,0] (kafka.server.ReplicaFetcherManager)
[2013-07-16 10:57:21,196] WARN [KafkaApi-3] Produce request with correlation id
2 from client on partition [dadj1,0] failed due to Leader not local for
partition [dadj1,0] on broker 3 (kafka.server.KafkaApis)
-----
dan@linux-rr29:~/git-data/kafka-current-src> bin/kafka-topics.sh --zookeeper
dev003:2181 --describe --topic dadj1 2>/dev/null
dadj1
configs:
partitions: 1
topic: dadj1 partition: 0 leader: 3 replicas: 3,4
isr: 3,4
dan@linux-rr29:~/git-data/kafka-current-src>
Dev003 logs show that server is elected as leader and has correct id of 3,
zookeeper shows dev003 is leader, but when I try to produce to the topic I get
a failure because the server thinks it is not the leader. This occurs
regardless of which server (dev003 or dev004) ends up the leader.
Here is my config which is the same except for the broker id and host names
[root@dev003 kafka-current-src]# grep -v -e '^#' -e '^$'
config/server.properties
broker.id=3
port=9092
host.name=dev003
num.network.threads=2
num.io.threads=2
socket.send.buffer.bytes=1048576
socket.receive.buffer.bytes=1048576
socket.request.max.bytes=104857600
log.dir=/opt/kafka/data/8.0/
num.partitions=1
log.flush.interval.messages=10000
log.flush.interval.ms=1000
log.retention.hours=168
log.segment.bytes=536870912
log.cleanup.interval.mins=1
zookeeper.connect=10.200.8.61:2181,10.200.8.62:2181,10.200.8.63:2181
zookeeper.connection.timeout.ms=1000000
kafka.metrics.polling.interval.secs=5
kafka.metrics.reporters=kafka.metrics.KafkaCSVMetricsReporter
kafka.csv.metrics.dir=/tmp/kafka_metrics
kafka.csv.metrics.reporter.enabled=false
[root@dev003 kafka-current-src]#
--
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
