使用Docker Compose部署多节点KRaft模式Kafka失败：元数据文件写入错误

尝试用Docker Compose部署采用KRaft共识协议的多节点Kafka集群，每个节点（kafka-1、kafka-2、kafka-3）均配置为controller+broker角色（KAFKA_CFG_PROCESS_ROLES=controller,broker），并定义了对应的listeners和advertised listeners。但启动时kafka-2和kafka-3出现元数据文件写入错误（Error while writing meta.properties file），导致节点以代码1退出，无法完成初始化。单节点（kafka-1）部署可正常运行，说明问题仅存在于多节点KRaft模式的Docker Compose配置中。

原配置文件 #

docker-compose.yml #

version: '3.8'

services:
  kafka-1:
    image: bitnami/kafka:latest
    user: 'root'
    ports: 
      - "9092:9092"
    environment:
      - KAFKA_CFG_NODE_ID=1
      - KAFKA_KRAFT_CLUSTER_ID=WnLkTHhk
      - KAFKA_CFG_PROCESS_ROLES=controller,broker
      - KAFKA_CFG_CONTROLLER_QUORUM_VOTERS=1@kafka-1:9091,2@kafka-2:9091,3@kafka-3:9091
      - KAFKA_CFG_LISTENERS=PLAINTEXT://:9090,CONTROLLER://:9091,EXTERNAL://:9092
      - KAFKA_CFG_ADVERTISED_LISTENERS=PLAINTEXT://kafka-1:9090,EXTERNAL://${HOSTNAME:-localhost}:9092
      - KAFKA_CFG_LISTENER_SECURITY_PROTOCOL_MAP=CONTROLLER:PLAINTEXT,EXTERNAL:PLAINTEXT,PLAINTEXT:PLAINTEXT
      - KAFKA_CFG_CONTROLLER_LISTENER_NAMES=CONTROLLER
      - KAFKA_CFG_INTER_BROKER_LISTENERS_NAME=PLAINTEXT
    volumes:
      - kafka-data:/bitnami/kafka

  kafka-2:
    image: bitnami/kafka:latest
    user: 'root'
    ports: 
      - "9094:9094"
    environment:
      - KAFKA_CFG_NODE_ID=2
      - KAFKA_KRAFT_CLUSTER_ID=WnLkTHhk
      - KAFKA_CFG_PROCESS_ROLES=controller,broker
      - KAFKA_CFG_CONTROLLER_QUORUM_VOTERS=1@kafka-1:9091,2@kafka-2:9091,3@kafka-3:9091
      - KAFKA_CFG_LISTENERS=PLAINTEXT://:9090,CONTROLLER://:9091,EXTERNAL://:9094
      - KAFKA_CFG_ADVERTISED_LISTENERS=PLAINTEXT://kafka-2:9090,EXTERNAL://${HOSTNAME:-localhost}:9094
      - KAFKA_CFG_LISTENER_SECURITY_PROTOCOL_MAP=CONTROLLER:PLAINTEXT,EXTERNAL:PLAINTEXT,PLAINTEXT:PLAINTEXT
      - KAFKA_CFG_CONTROLLER_LISTENER_NAMES=CONTROLLER
      - KAFKA_CFG_INTER_BROKER_LISTENERS_NAME=PLAINTEXT
    volumes:
      - kafka-data:/bitnami/kafka

  kafka-3:
    image: bitnami/kafka:latest
    user: 'root'
    ports: 
      - "9096:9096"
    environment:
      - KAFKA_CFG_NODE_ID=3
      - KAFKA_KRAFT_CLUSTER_ID=WnLkTHhk
      - KAFKA_CFG_PROCESS_ROLES=controller,broker
      - KAFKA_CFG_CONTROLLER_QUORUM_VOTERS=1@kafka-1:9091,2@kafka-2:9091,3@kafka-3:9091
      - KAFKA_CFG_LISTENERS=PLAINTEXT://:9090,CONTROLLER://:9091,EXTERNAL://:9096
      - KAFKA_CFG_ADVERTISED_LISTENERS=PLAINTEXT://kafka-3:9090,EXTERNAL://${HOSTNAME:-localhost}:9096
      - KAFKA_CFG_LISTENER_SECURITY_PROTOCOL_MAP=CONTROLLER:PLAINTEXT,EXTERNAL:PLAINTEXT,PLAINTEXT:PLAINTEXT
      - KAFKA_CFG_CONTROLLER_LISTENER_NAMES=CONTROLLER
      - KAFKA_CFG_INTER_BROKER_LISTENERS_NAME=PLAINTEXT
    volumes:
      - kafka-data:/bitnami/kafka

volumes:
  kafka-data:

environment.env #

HOSTNAME=host.docker.internal

错误日志 #

kafka-1_1  | kafka 22:33:12.35 INFO  ==> Formatting storage directories to add metadata...
kafka-3_1  | metaPropertiesEnsemble=MetaPropertiesEnsemble(metadataLogDir=Optional.empty, dirs={/bitnami/kafka/data: EMPTY})
kafka-3_1  | Formatting /bitnami/kafka/data with metadata.version 3.7-IV4.
kafka-2_1  | metaPropertiesEnsemble=MetaPropertiesEnsemble(metadataLogDir=Optional.empty, dirs={/bitnami/kafka/data: EMPTY})
kafka-2_1  | Formatting /bitnami/kafka/data with metadata.version 3.7-IV4.
kafka-1_1  | metaPropertiesEnsemble=MetaPropertiesEnsemble(metadataLogDir=Optional.empty, dirs={/bitnami/kafka/data: EMPTY})
kafka-1_1  | Formatting /bitnami/kafka/data with metadata.version 3.7-IV4.
kafka-3_1  | Error while writing meta.properties file /bitnami/kafka/data: /bitnami/kafka/data/bootstrap.checkpoint.tmp
kafka-2_1  | Error while writing meta.properties file /bitnami/kafka/data: /bitnami/kafka/data/bootstrap.checkpoint.tmp
kafka-1_1  | kafka 22:33:14.22 INFO  ==> ** Kafka setup finished! **
kafka-1_1  | 
kafka-1_1  | kafka 22:33:14.24 INFO  ==> ** Starting Kafka **
architecture-experiments_kafka-2_1 exited with code 1
architecture-experiments_kafka-3_1 exited with code 1

执行命令 #

docker-compose --env-file environment.env up

问题原因 #

三个Kafka节点共享了同一个kafka-data卷，导致多个节点同时向同一个文件路径写入元数据文件，引发文件锁冲突或写入覆盖，这是kafka-2和kafka-3写入失败的直接原因。单节点部署时没有竞争，因此可以正常运行。

解决方案 #

为每个Kafka节点分配独立的存储卷，避免文件写入冲突。修改后的配置如下：

修改后的docker-compose.yml #

version: '3.8'

services:
  kafka-1:
    image: bitnami/kafka:latest
    user: 'root'
    ports: 
      - "9092:9092"
    environment:
      - KAFKA_CFG_NODE_ID=1
      - KAFKA_KRAFT_CLUSTER_ID=WnLkTHhk
      - KAFKA_CFG_PROCESS_ROLES=controller,broker
      - KAFKA_CFG_CONTROLLER_QUORUM_VOTERS=1@kafka-1:9091,2@kafka-2:9091,3@kafka-3:9091
      - KAFKA_CFG_LISTENERS=PLAINTEXT://:9090,CONTROLLER://:9091,EXTERNAL://:9092
      - KAFKA_CFG_ADVERTISED_LISTENERS=PLAINTEXT://kafka-1:9090,EXTERNAL://${HOSTNAME:-localhost}:9092
      - KAFKA_CFG_LISTENER_SECURITY_PROTOCOL_MAP=CONTROLLER:PLAINTEXT,EXTERNAL:PLAINTEXT,PLAINTEXT:PLAINTEXT
      - KAFKA_CFG_CONTROLLER_LISTENER_NAMES=CONTROLLER
      - KAFKA_CFG_INTER_BROKER_LISTENERS_NAME=PLAINTEXT
    volumes:
      - kafka-data-1:/bitnami/kafka

  kafka-2:
    image: bitnami/kafka:latest
    user: 'root'
    ports: 
      - "9094:9094"
    environment:
      - KAFKA_CFG_NODE_ID=2
      - KAFKA_KRAFT_CLUSTER_ID=WnLkTHhk
      - KAFKA_CFG_PROCESS_ROLES=controller,broker
      - KAFKA_CFG_CONTROLLER_QUORUM_VOTERS=1@kafka-1:9091,2@kafka-2:9091,3@kafka-3:9091
      - KAFKA_CFG_LISTENERS=PLAINTEXT://:9090,CONTROLLER://:9091,EXTERNAL://:9094
      - KAFKA_CFG_ADVERTISED_LISTENERS=PLAINTEXT://kafka-2:9090,EXTERNAL://${HOSTNAME:-localhost}:9094
      - KAFKA_CFG_LISTENER_SECURITY_PROTOCOL_MAP=CONTROLLER:PLAINTEXT,EXTERNAL:PLAINTEXT,PLAINTEXT:PLAINTEXT
      - KAFKA_CFG_CONTROLLER_LISTENER_NAMES=CONTROLLER
      - KAFKA_CFG_INTER_BROKER_LISTENERS_NAME=PLAINTEXT
    volumes:
      - kafka-data-2:/bitnami/kafka

  kafka-3:
    image: bitnami/kafka:latest
    user: 'root'
    ports: 
      - "9096:9096"
    environment:
      - KAFKA_CFG_NODE_ID=3
      - KAFKA_KRAFT_CLUSTER_ID=WnLkTHhk
      - KAFKA_CFG_PROCESS_ROLES=controller,broker
      - KAFKA_CFG_CONTROLLER_QUORUM_VOTERS=1@kafka-1:9091,2@kafka-2:9091,3@kafka-3:9091
      - KAFKA_CFG_LISTENERS=PLAINTEXT://:9090,CONTROLLER://:9091,EXTERNAL://:9096
      - KAFKA_CFG_ADVERTISED_LISTENERS=PLAINTEXT://kafka-3:9090,EXTERNAL://${HOSTNAME:-localhost}:9096
      - KAFKA_CFG_LISTENER_SECURITY_PROTOCOL_MAP=CONTROLLER:PLAINTEXT,EXTERNAL:PLAINTEXT,PLAINTEXT:PLAINTEXT
      - KAFKA_CFG_CONTROLLER_LISTENER_NAMES=CONTROLLER
      - KAFKA_CFG_INTER_BROKER_LISTENERS_NAME=PLAINTEXT
    volumes:
      - kafka-data-3:/bitnami/kafka

volumes:
  kafka-data-1:
  kafka-data-2:
  kafka-data-3:

操作步骤 #

1. 停止并删除现有容器和共享卷：

docker-compose down -v

2. 使用修改后的配置重新启动集群：

docker-compose --env-file environment.env up

内容的提问来源于stack exchange，提问作者Mateus Levi