Spark集群启动时worker节点启不起来
在spark集群中使用命令:sbin/start-all.sh启动集群时报错:starting org.apache.spark.deploy.master.Master, logging to /home/yxk/cluster/spark/logs/spark-yxk-org.apache.spark.deploy.master.Master-1-linux.outyxk@l...
·
在spark集群中使用命令:
sbin/start-all.sh
启动集群时报错:
starting org.apache.spark.deploy.master.Master, logging to /home/yxk/cluster/spark/logs/spark-yxk-org.apache.spark.deploy.master.Master-1-linux.out
yxk@linux's password:
linux: starting org.apache.spark.deploy.worker.Worker, logging to /home/yxk/cluster/spark/logs/spark-yxk-org.apache.spark.deploy.worker.Worker-1-linux.out
linux: failed to launch: nice -n 0 /home/yxk/cluster/spark/bin/spark-class org.apache.spark.deploy.worker.Worker --webui-port 8081 spark://linux:7077
linux: JAVA_HOME is not set
linux: full log in /home/yxk/cluster/spark/logs/spark-yxk-org.apache.spark.deploy.worker.Worker-1-linux.out
从日志分析中可以看出启动时找不到JAVA_HOME
不过检查当前用户下的java环境是正常的,经分析寻找资料在这个博客中找到了 原理https://stackoverflow.com/questions/33955635/why-does-start-all-sh-from-root-cause-failed-to-launch-org-apache-spark-deploy
解决方法
在当前用户的.bashrc文件中加入java的环境变量
$ vim ~/.bashrc
加入java环境变量
重新启动spark集群
使用jps命令查看当前的进程。显示正常
更多推荐
已为社区贡献1条内容
所有评论(0)