SpringBoot集成Kafka快速启动示例
源码地址: https://github.com/ghdefe/kafka-demo
kafka消息发送接收示例
1. 启动kafka实例
cat << EOF > docker-compose.yml
version: "2"
services:
zookeeper:
image: docker.io/bitnami/zookeeper:3.8
ports:
- "2181:2181"
- "2180:8080"
environment:
- ALLOW_ANONYMOUS_LOGIN=yes
kafka:
image: docker.io/bitnami/kafka:3.2
ports:
- "9092:9092"
environment:
- KAFKA_CFG_ZOOKEEPER_CONNECT=zookeeper:2181
- ALLOW_PLAINTEXT_LISTENER=yes
- KAFKA_CFG_MAX_REQUEST_SIZE=1195725856
- KAFKA_CFG_LISTENERS=PLAINTEXT://:9092
- KAFKA_CFG_ADVERTISED_LISTENERS=PLAINTEXT://localhost:9092
depends_on:
- zookeeper
EOF
docker-compose up -d
2. 启动消费者
消费者主要代码: com/example/consumer/service/MsgConsumer.java
, 消费者启动类: com/example/consumer/ConsumerApplication.java
.
测试消息均衡分配效果需要启动多个消费者实例,因此在IntelliJ Idea中右上角, 启动配置中配置Allow multiple instances
, 就可以多次点击启动按钮启动多个消费者实例.
3. 启动生产者
生产者配置: com/example/producer/config/TopicConfiguration.java
, 在这里定义了multi-topic
主题的分区数为4,用于测试消息均衡效果.
生产消息: 运行此类可发送1条消息到single-topic
和1000条消息到multi-topic
. com/example/producer/service/MsgProducerTest.java
笔记
消费者组中的成员会自动被分配分区, 消费组中的成员自动
且均衡
地分配所有的分区. 例如:
multi-topic
有4个分区.
- 启动第一个消费者实例时, 会打印以下内容:
2024-07-08T15:46:19.578+08:00 INFO 21836 --- [consumer] [ntainer#2-0-C-1] k.c.c.i.ConsumerRebalanceListenerInvoker : [Consumer clientId=consumer-consumer-0, groupId=consumer-group2] Adding newly assigned partitions: multi-topic-0, multi-topic-1, multi-topic-2, multi-topic-3
2024-07-08T15:46:19.604+08:00 INFO 21836 --- [consumer] [ntainer#2-0-C-1] o.s.k.l.KafkaMessageListenerContainer : consumer-group2: partitions assigned: [multi-topic-0, multi-topic-1, multi-topic-2, multi-topic-3]
可见consumer-group2
分配了4个分区: multi-topic-0, multi-topic-1, multi-topic-2, multi-topic-3
.
- 接着启动第二个消费者实例
第一个消费者打印日志
2024-07-08T15:51:42.526+08:00 INFO 21836 --- [consumer] [ntainer#2-0-C-1] o.s.k.l.KafkaMessageListenerContainer : consumer-group2: partitions revoked: [multi-topic-0, multi-topic-1, multi-topic-2, multi-topic-3]
2024-07-08T15:51:42.602+08:00 INFO 21836 --- [consumer] [ntainer#2-0-C-1] k.c.c.i.ConsumerRebalanceListenerInvoker : [Consumer clientId=consumer-consumer-0, groupId=consumer-group2] Adding newly assigned partitions: multi-topic-0, multi-topic-1
2024-07-08T15:51:42.607+08:00 INFO 21836 --- [consumer] [ntainer#2-0-C-1] o.s.k.l.KafkaMessageListenerContainer : consumer-group2: partitions assigned: [multi-topic-0, multi-topic-1]
第二个消费者打印日志
2024-07-08T15:51:42.602+08:00 INFO 21276 --- [consumer] [ntainer#1-0-C-1] k.c.c.i.ConsumerRebalanceListenerInvoker : [Consumer clientId=consumer-consumer-0, groupId=consumer-group2] Adding newly assigned partitions: multi-topic-2, multi-topic-3
2024-07-08T15:51:42.612+08:00 INFO 21276 --- [consumer] [ntainer#1-0-C-1] o.s.k.l.KafkaMessageListenerContainer : consumer-group2: partitions assigned: [multi-topic-2, multi-topic-3]
可见在第二个消费者启动时, 会触发分区重新分配
- 第一个消费者先撤销现有4个分区分配, 然后重新获取分配的分区, 最终绑定分区: multi-topic-0, multi-topic-1
- 第二个消费者获取分配的分区, 最终绑定分区: multi-topic-2, multi-topic-3
可见: 消费组中的成员自动
且均衡
地分配所有的分区.