kafka2.6.0 installation configuration

Posted by juschillinnow on Thu, 10 Feb 2022 13:58:48 +0100

kafka installation record:

Official website: http://kafka.apache.org/downloads.html

1. Download

yum install -y wget
wget https://mirrors.bfsu.edu.cn/apache/kafka/2.6.0/kafka_2.12-2.6.0.tgz 

2. Decompression:

tar -zxvf kafka_2.12-2.6.0.tgz -C /opt/

3. Change of name

mv kafka_2.12-2.6.0 kafka

4. Create files

cd kafka

Create under kafka Directory:

mkdir logs 
mkdir data

5. Modify the configuration file:

cd config/

vi server.properties

#The globally unique number of the broker, which cannot be repeated
#Delete topic function enable
#Number of threads processing network requests
#Off the shelf quantity used to process disk IO
#Buffer size of send socket
#Buffer size of receive socket
#Buffer size of the requested socket
#kafka running log storage path
#Number of partitions of topic on the current broker
#Number of threads used to recover and clean up data under data
#The maximum time the segment file is retained. If the timeout expires, it will be deleted
#Configure Zookeeper cluster address

Modify the configuration file on the other two hosts respectively
/opt/kafka/config/server. Broker in properties id=1,broker.id=2

Note: broker ID cannot be repeated

6. Configure environment variables

vi ~/.bash_profile

add to:


export KAFKA_HOME=/opt/kafka


Profile validation:

source ~/.bash_profile

7. Start the cluster,

(you must start zookeeper first) (write a startup script later)

cd /opt/zookeeper/bin/
./zkServer.sh start

7.1 start kafka on the master, server1 and server2 nodes in turn

cd /opt/kafka/bin/

./kafka-server-start.sh /opt/kafka/config/server.properties

(after startup, it is a blocking process, and the operation needs to reopen a window, but it is convenient to view the output log)

./kafka-server-start.sh -daemon /opt/kafka/config/server.properties

(background startup, recommended.)

Check whether Kafka starts successfully jps

8. Create cluster script

echo $PATH

In order to find the script we created; It can be created in any of the above bin directories; Take / usr/bin as an example

cd /usr/bin 
vi kk.sh
8.1 change kafka stop script

kafka server stop. In the config directory of kafka SH has some problems. You need to modify the stop script provided by the official first. Every machine in the cluster should be changed (note that k in kafka is lowercase)

PIDS=$(ps ax | grep -i 'kafka\.Kafka' | grep java | grep -v grep | awk '{print $1}')

Change to

PIDS=$(ps ax | grep -i 'kafka' | grep java | grep -v grep | awk '{print $1}')

Cluster startup script:


case $i in
	for i in x x1 x2
			echo --------$i start-up kafka---------
			ssh $i " source /etc/profile;/opt/kafka/bin/kafka-server-start.sh -daemon /opt/kafka/config/server.properties"
	for i in x x1 x2
			echo --------$i stop it kafka---------
			ssh $i " source /etc/profile;/opt/kafka/bin/kafka-server-stop.sh stop"
8.2 authorization:
chmod 777 kk.sh
8.3 script usage:

Stop: KK sh stop

Operation: KK sh start

Solve the problem that script files created by Windows cannot be uploaded to Linux:

sed -i 's/\r$//' kk.sh

9. Create theme:

[root@x kafka]#

bin/kafka-topics.sh --create --zookeeper x:2181 --topic first --partitions 2 --replication-factor 2 --create

bin/kafka-topics.sh added theme create to create zookeeper x:2181
Storage data topic first topic name partitions number of partitions replication factor 2 copies

9.1 viewing topic:
[root@x kafka]# bin/kafka-topics.sh --list --zookeeper x:2181

Help us classify data

9.2 delete subject:
bin/kafka-topics.sh --delete --zookeeper x:2181 --topic first 

Server is required Set delete in properties topic. Enable = true, otherwise it is only marked for deletion.

9.3 modify the number of partitions
bin/kafka-topics.sh --zookeeper x:2181 --alter --topic first --partitions 3

The number of general partitions shall not exceed the number of hosts

9.4 details:
bin/kafka-topics.sh --describe --topic first --zookeeper x:2181

10. Production:

bin/kafka-console-producer.sh --topic first --broker-list x:9092

10.1 turn on another machine:

[root@x1 kafka]# bin/kafka-console-consumer.sh --bootstrap-server x:9092,x1:9092,x2:9092 --topic first

x producers send messages, x1 consumers receive messages,

x2 consumers can't spend more than 7 days after receiving messages from the beginning

bin/kafka-console-consumer.sh --bootstrap-server x:9092,x1:9092,x2:9092 --topic first --from-beginning

In case of the following warnings, a node is disconnected. Check the status of three nodes. If it is disconnected, it can be solved after the node is restarted

11 some error aggregation records

zookeeper, kafka error
1.ERROR Exiting Kafka due to fatal exception (kafka.Kafka$)

Cause: not found properties File or configuration file path specified incorrectly
 Method: specify the configuration file path

2.Configured broker.id 130 doesn't match stored broker.id 0 in meta.properties

Reason: single machine used
 Method: delete the data file and restart

3.zookeeper is not a recognized option

Reason: old version used--zookeeper This option has been removed from the new version

Method: 0.9 Later versions start with --bootstrap-server

4.Replication factor: 2 larger than available brokers: 1

Reason: the number of copies exceeded broker Number of servers
 Method: less than broker Just count

5. Error reporting:

WARN Error while fetching metadata with correlation id 479 : {test=LEADER_NOT_AVAILABLE} (org.apache.kafka.clients.NetworkClient)

WARN Error while fetching metadata with correlation id 480 : {test=LEADER_NOT_AVAILABLE} (org.apache.kafka.clients.NetworkClient)

Method: modify server.properties add to

Topics: Linux Big Data Hadoop kafka Zookeeper