apache-spark - Spark 2.0 状态 : COMPLETE Exit status code -100 on yarn

Question

有人可以指点我有关 -100 退出代码含义的文档吗？ EMR 集群，YARN 上的 spark 2.0.0(根据 EMR 标准 spark-cluster 部署)。我见过https://www.cloudera.com/documentation/enterprise/5-8-x/topics/cdh_sg_yarn_container_exec_errors.html这给出了一些错误代码，其中 -100 不是其中之一。此外，作为一个更普遍的问题，YARN 容器日志和 Spark 容器日志似乎都没有包含很多关于导致此类故障的原因的信息......从我看到的 YARN 日志中
17/01/18 17:51:58 INFO YarnAllocator: Canceling requests for 1 executor container(s) to have a new desired total 4164 executors.
17/01/18 17:51:58 INFO YarnAllocator: Driver requested a total number of 4163 executor(s).
17/01/18 17:51:58 INFO YarnAllocator: Canceling requests for 1 executor container(s) to have a new desired total 4163 executors.
17/01/18 17:51:58 INFO YarnAllocator: Driver requested a total number of 4162 executor(s).
17/01/18 17:51:58 INFO YarnAllocator: Canceling requests for 1 executor container(s) to have a new desired total 4162 executors.
17/01/18 17:51:59 INFO YarnAllocator: Driver requested a total number of 4161 executor(s).
17/01/18 17:51:59 INFO YarnAllocator: Driver requested a total number of 4160 executor(s).
17/01/18 17:51:59 INFO YarnAllocator: Canceling requests for 2 executor container(s) to have a new desired total 4160 executors.
17/01/18 17:52:00 INFO YarnAllocator: Driver requested a total number of 4159 executor(s).
17/01/18 17:52:00 INFO YarnAllocator: Canceling requests for 1 executor container(s) to have a new desired total 4159 executors.
17/01/18 17:52:00 INFO YarnAllocator: Completed container container_1483555419510_0037_01_000114 on host: ip-172-20-221-152.us-west-2.compute.internal (state: COMPLETE, exit status: -100)
17/01/18 17:52:00 WARN YarnAllocator: Container marked as failed: container_1483555419510_0037_01_000114 on host: ip-172-20-221-152.us-west-2.compute.internal. Exit status: -100. Diagnostics: Container released on a *lost* node
17/01/18 17:52:00 INFO YarnAllocator: Completed container container_1483555419510_0037_01_000107 on host: ip-172-20-221-152.us-west-2.compute.internal (state: COMPLETE, exit status: -100)
17/01/18 17:52:00 WARN YarnAllocator: Container marked as failed: container_1483555419510_0037_01_000107 on host: ip-172-20-221-152.us-west-2.compute.internal. Exit status: -100. Diagnostics: Container released on a *lost* node
17/01/18 17:52:00 INFO YarnAllocator: Will request 2 executor containers, each with 7 cores and 22528 MB memory including 2048 MB overhead
17/01/18 17:52:00 INFO YarnAllocator: Canceled 0 container requests (locality no longer needed)
17/01/18 17:52:00 INFO YarnAllocator: Submitted container request (host: Any, capability: <memory:22528, vCores:7>)
17/01/18 17:52:00 INFO YarnAllocator: Submitted container request (host: Any, capability: <memory:22528, vCores:7>)
17/01/18 17:52:01 INFO YarnAllocator: Driver requested a total number of 4158 executor(s).
17/01/18 17:52:01 INFO YarnAllocator: Canceling requests for 1 executor container(s) to have a new desired total 4158 executors.
17/01/18 17:52:02 INFO YarnAllocator: Driver requested a total number of 4157 executor(s).
和我看到的 Spark 执行器日志
17/01/18 17:39:39 INFO MemoryStore: MemoryStore cleared
17/01/18 17:39:39 INFO BlockManager: BlockManager stopped
17/01/18 17:39:39 INFO ShutdownHookManager: Shutdown hook called
哪一个都不是很有用？

Answer 1

“退出状态:-100。诊断:容器在丢失的节点上发布”告诉您该节点已丢失