Fang Jin Ask A question, if schedulis and dss are installed on different machines, dss stand-alone version is installed on machine A (including hadoop\hive\spark), and schedulis is installed on machine B, then when configuring azkaban.properties of schedulis, hadoop.home= hadoop.conf.dir= hive.home= spark.home= How d are these values specified?
Zhang Huajin The normal practice is to make a copy of the schedule node Fang Jin That's two separate sets of hadoop\hive\spark, right? Zhang Huajin If not, it is recommended to learn related concepts Fang Jin Do you want to build a hadoop cluster, or just copy the hadoop related directory to the past? Zhang Huajin for reference only CDH5.16.1 cluster enterprises are truly deployed offline https://mp.weixin.qq.com/s/65y28Fs61IrOehHkxpWxoA In general, all deployed linkis or scheduled machines must be gateway nodes. New deployment gateway can refer to this How do I configure a Gateway node for a Kerberos environment outside of a CDH cluster https://mp.weixin.qq.com/s/Y6iirtXuamJGtIzllKX2Og Fang Jin Ok, thank you. Fang Jin 问个问题，如果schedulis与dss装在不同机器上，dss单机版先装在A机器上（包括hadoop\hive\spark), schedulis装在B机器上，那在配置schedulis的azkaban.properties时， hadoop.home= hadoop.conf.dir= hive.home= spark.home= 这几个值怎么指定 ？？ Zhang Huajin 正常做法是拷贝一份放调度节点 Fang Jin 拷贝一份那不是两套独立的hadoop\hive\spark了吗？ Zhang Huajin 不是的 建议学习一下相关概念 Fang Jin 是要建一个hadoop集群，还是把hadoop相关的目录拷过去就行了，具体说说呢？ Zhang Huajin 供参考 CDH5.16.1集群企业真正离线部署 https://mp.weixin.qq.com/s/65y28Fs61IrOehHkxpWxoA 一般所有的部署linkis或者调度的机器必须是gateway 节点 新增部署gateway 可以参考这个 如何在CDH集群外配置Kerberos环境的Gateway节点 https://mp.weixin.qq.com/s/Y6iirtXuamJGtIzllKX2Og Fang Jin 好的，谢谢
