在每个应用程序上重新读取哪些YARN配置参数？(Which YARN configuration parameters are re

在每个应用程序上重新读取哪些YARN配置参数？(Which YARN configuration parameters are re-read on each application?)

我有一份工作比我在日常工作流程中运行的其他大约50个工作要大得多。 我希望物业yarn.app.mapreduce.am.resource.mb对于大工作来说更大。 我好运吗？ 如何判断哪些属性需要完全重新启动才能更改我的群集中的哪些YARN？

I've got one job that's much bigger than the other 50 or so that run in my daily workflow. I'd like the property yarn.app.mapreduce.am.resource.mb to be larger for just the big job. Am I in luck? How can I tell which properties require a complete restart of which pieces of YARN across my cluster to be changed?

最满意答案

可以在作业级别更改此设置，而无需重新启动YARN服务。 一般情况下，您可以更改每个MR属性。 通常，作业级属性（如mapper / reducer / appmaster内存设置）可以在作业级别更新。

您可以查看此有用的帖子 ，以深入了解此主题。

This setting can be changed on job level without YARN service restart. It's hard to tell in general when you can change each MR property. Usually job level properties like mapper/reducer/appmaster memory settings can be updated on job level.

You can check this helpful post to get a deeper look on this topic.