kafka topic 的leader为none问题处理

原创于 2025-04-17 17:29:35 发布 · 146 阅读

0 ·

CC 4.0 BY-SA版权

文章标签：

#kafka #分布式

查看为broker主节点的日志，报错：
ERROR state.change.logger: [Controller id=322 epoch=47] Controller 322 epoch 47 failed to change state for partition proj-gps-30-md-measuredata-j01-3-2 from NewPartition to NewPartition
org.apache.zookeeper.KeeperException$SessionMovedException: KeeperErrorCode = Session moved for /controller_epoch
at org.apache.zookeeper.KeeperException.create(KeeperException.java:134)
排查会话参数：
ZooKeeper maxSessionTimeout 60000 最大会话超时时间（毫秒）
Kafka zookeeper.session.timeout.ms 30000 Kafka 与 ZooKeeper 会话超时时间
Kafka controlled.shutdown.enable true 确保 Broker 关闭前迁移 Controller
修改zookeeper.session.timeout.ms 为60s，重启broker后恢复。