[英]Is it OK to force the hosting of the applicationMaster on one same node (YARN)?
I am submitting Spark applications to my Hadoop 3 nodes cluster.我正在将 Spark 应用程序提交到我的 Hadoop 3 节点集群。 In my
spark-defaults.conf
file i set在我设置的
spark-defaults.conf
文件中
spark.yarn.appMasterEnv.SPARK_LOCAL_IP 127.0.0.1
spark.yarn.appMasterEnv.SPARK_MASTER_HOST 0.0.0.0
So that the applicationMaster is always (client or cluster mode) hosted on the client machine.这样 applicationMaster 总是(客户端或集群模式)托管在客户端机器上。 Is it OK to do that ?
这样做可以吗?
To note that if i don't do that and Yarn attempts to host the applicationMaster on a slave node then binding error stops the running.请注意,如果我不这样做并且 Yarn 尝试在从属节点上托管 applicationMaster,则绑定错误会停止运行。
Thanks for clarifying this.感谢您澄清这一点。
If this is just for you and this works, go for it.如果这仅适合您并且有效,那就去做吧。
You aren't following a "Normal" spark strategy for a yarn cluster.您没有遵循纱线簇的“正常”火花策略。 Is that 'OK'?
这可以吗'? If you have a good reason, yes it's ok.
如果你有充分的理由,是的,没关系。
Would I use this in production?我会在生产中使用它吗? No.
不。
Are there simpler more common ways of running a cluster?是否有更简单更常见的集群运行方式? Yes.
是的。
You are mixing strategies of running Spark Standalone and Yarn.您正在混合运行 Spark Standalone 和 Yarn 的策略。 These are two fundamentally different architectures.
这是两种根本不同的架构。 If you can make the two architectures work together that's fun.
如果你能让这两种架构一起工作,那就很有趣了。 But you may hit some weird problems and as this is a custom set of settings you may not find a lot of support to help you.
但是您可能会遇到一些奇怪的问题,而且由于这是一组自定义设置,您可能找不到很多支持来帮助您。
No, It's not "OK".不,这不是“好”。
One of the ideologies behind spark is resilience. spark背后的意识形态之一是弹性。 If you are forcing 1 node to be the application master you are introducing a bottleneck & a single point of failure.
如果您强制 1 个节点成为应用程序主节点,则会引入瓶颈和单点故障。 You are using yarn, there is no reason to specify a master.
您正在使用纱线,没有理由指定主人。
声明:本站的技术帖子网页,遵循CC BY-SA 4.0协议,如果您需要转载,请注明本站网址或者原文地址。任何问题请咨询:yoyou2525@163.com.