【发布时间】:2026-01-05 05:20:12
【问题描述】:
我正在运行一个由三台机器组成的 Zookeeper ensemble,一台 Mesos 主机作为领导者,另外两台作为追随者。我已经看到,如果我故意让当前的 Mesos 主人失败/停止,那么其他两个追随者中的任何一个都将被选为 Mesos 主人。这是我理解的预期行为。
My question is, if I submit a job via a framework and during the job execution if the current Mesos master fails, then how will the remaining tasks in that job be handled by the next elected Mesos leader?
For example, say during the execution of the sparkPi example if the current Mesos master fails then how do I tell the next elected master to pick up the tasks from where the last master left them?这个故障转移是由 zookeeper 自动处理还是需要任何配置更改。
【问题讨论】:
标签: apache-spark apache-zookeeper high-availability failover