Prerequesites
Install three nodes with CentOS 7 with at least 20GB Disk, 2 GB RAM and two CPU Cores.
Install JDK
yum install -y java-1.8.0-openjdkl java-1.8.0-openjdk-devel net-tools
Set JAVA_HOME in ~/.bashrc
# Set Java-Home export JAVA_HOME="/usr/lib/jvm/java-1.8.0-openjdk-1.8.0.151-5.b12.el7_4.x86_64/jre" export PATH=$JAVA_HOME/bin:$PATH
Disable SELinux, Firewall and IPv6
systemctl disable firewalld systemctl stop firewalld echo "net.ipv6.conf.all.disable_ipv6 = 1" >> /etc/sysctl.conf [root@kafka3 ~]# cat /etc/selinux/config | grep "^SELINUX=" SELINUX=permissive
Reboot Server
Installing Kafka
Download Kafka and unpack it under /opt
https://www.apache.org/dyn/closer.cgi?path=/kafka/0.11.0.2/kafka_2.11-0.11.0.2.tgz
tar zxvf kafka_2.11-0.11.0.2.tgz
Starting Zookeeper
On each node create a zookeeper directory and a file ‚myid‘ with a unique number:
mkdir /zookeeper echo '1' > /zookeeper/myid
On all three Server go to Kafka home folder /opt/kafka_2.11-0.11.0.1 and setup zookeeper like this
vi config/zookeeper.properties
# the directory where the snapshot is stored. dataDir=/zookeeper # the port at which the clients will connect clientPort=2181 clientPortAddress=192.168.2.56 # disable the per-ip limit on the number of connections since this is a non-production config maxClientCnxns=0 # The number of milliseconds of each tick tickTime=2000 # The number of ticks that the initial synchronization phase can take initLimit=10 # The number of ticks that can pass between # sending a request and getting an acknowledgement syncLimit=5 # zoo servers server.1=kafka1.fritz.box:2888:3888 server.2=kafka2.fritz.box:2888:3888 server.3=kafka3.fritz.box:2888:3888 #add here more servers if you want
Start Zookeeper on all three servers
./bin/zookeeper-server-start.sh -daemon config/zookeeper.properties
Change the Kafka server.properties on all three servers (set a unique broker id on each server)
vi config/server.properties
# The id of the broker. This must be set to a unique integer for each broker. broker.id=2 # listeners = PLAINTEXT://your.host.name:9092 listeners=PLAINTEXT://:9093 # A comma seperated list of directories under which to store log files log.dirs=/tmp/kafka-logs-2 # Zookeeper connection string (see zookeeper docs for details). # This is a comma separated host:port pairs, each corresponding to a zk # server. e.g. "127.0.0.1:3000,127.0.0.1:3001,127.0.0.1:3002". # You can also append an optional chroot string to the urls to specify the # root directory for all kafka znodes. zookeeper.connect=kafka1.fritz.box:2181,kafka2.fritz.box:2181,kafka3.fritz.box
Start Kafka on all three nodes:
./bin/kafka-server-start.sh -daemon config/server.properties
Verify kafka and zookeper are running:
jps 4150 Jps 2365 QuorumPeerMain 1743 Kafka
Verify also all brokers are registered to zookeeper:
# ./bin/zookeeper-shell.sh kafka1:2181 ls /brokers/ids Connecting to kafka1:2181 WATCHER:: WatchedEvent state:SyncConnected type:None path:null [1, 2, 3]
Create a example Topic with three partitions and replicationfactor 3
# ./bin/kafka-topics.sh --create --zookeeper kafka1:2181 --topic example-topic --partitions 3 --replication-factor 3 Created topic "example-topic". # ./bin/kafka-topics.sh --list --zookeeper kafka1:2181 --topic example-topic example-topic # ./bin/kafka-topics.sh --describe --zookeeper kafka1:2181 --topic example-topic Topic:example-topic PartitionCount:3 ReplicationFactor:3 Configs: Topic: example-topic Partition: 0 Leader: 2 Replicas: 2,3,1 Isr: 2,3,1 Topic: example-topic Partition: 1 Leader: 3 Replicas: 3,1,2 Isr: 3,2,1 Topic: example-topic Partition: 2 Leader: 1 Replicas: 1,2,3 Isr: 1,2,3
Test the Topic
Start a Producer on one node:
# ./bin/kafka-console-producer.sh --broker-list kafka1:9093,kafka2:9093,kafka3:9093 --topic example-topic
Start also a Consumer on a different node:
# ./bin/kafka-console-consumer.sh --zookeeper kafka1:2181 --topic example-topic --from-beginning
Write some text in the producer console. You should then see the Text on the Consumer Console.
Stop a node and write again some messages in the producer console to verify the high availability is working.