Hi, This issue happens when your zookeeper cluster is down , in that case the znode is not removed.
We are running both zookeeper and kafka at same machine , hence if the machine goes down causing <= 50% node in zookeeper cluster, there is no way currently to clear the /broker/ids/ nodes. Also i am not sure if same /broker/ids/ gets deleted as soon as kafka is killed. Thanks Harish On Fri, Jun 8, 2018 at 10:51 PM 逐风者的祝福 <1878707...@qq.com> wrote: > I think /broker/ids/ is registered as a temporary znode when the broker is > down and znode will be removed . > > > > > ------------------ 原始邮件 ------------------ > 发件人: "harish lohar"<hklo...@gmail.com>; > 发送时间: 2018年6月7日(星期四) 晚上11:41 > 收件人: "users"<users@kafka.apache.org>; > 主题: kafka Broker Znode TTL > > > > Do Kafka Broker has configuration to specify TTL for kafka broker znodes > such /broker/ids/ > > This is useful in case zookeeper cluster goes down and kafka broker > re-registers , in that case it fails to start as Zookeeper already has the > same broker id registered. > > Thanks > Harish
