问题描述
我刚刚安装了apache zeppelin(从git repo的最新资源构建)，并成功地看到它在端口10008中启动并运行.我用一行代码创建了一个新的笔记本
I just installed apache zeppelin (built from latest source from git repo) and successfully saw it is up and running in the port 10008. I created a new note book with a single line of code
val a = "Hello World!"
运行本段并看到以下错误
And run this paragraph and saw the below error
有任何线索吗?
我的后端是spark 1.5，并且我通过解释器的Web界面进行了验证，齐柏林飞艇指向了正确版本的spark和适当的spark.home.
My backend is spark 1.5 and I verified by web interface of interpreter that zeppelin points to right version of spark and approproate spark.home.
推荐答案
在Zeppelin尝试创建解释器时发生的错误也可能导致该错误.
The error may be casued also du to an error occurred while Zeppelin trying to create the interpreter.
Zeppelin在另一个过程中启动了解释程序，并尝试使用Thrift协议进行连接
Zeppelin starts the interpretter in a different process and tries to connect to using Thrift Protocol
在我的情况下，尝试在spark-defaults.conf中为Spark驱动程序分配5GB时遇到此错误注释此行(或分配为4g或更少)时可以解决
In my case I ve this error when trying to assign 5GB for spark driver in spark-defaults.confIt is resolved when commenting this line (or assign 4g or less)
#spark.driver.memory 5g
您可以查看此JIRA ZEPPELIN-305
You could have a look at this JIRA ZEPPELIN-305
编辑:
此错误可能是由于阻止Spark解释器进程启动的任何原因引起的.最近，当尝试将JMX选项添加到ZEPPELIN_JAVA_OPTS时，我得到了它，这导致解释器进程使用与Zeppelin进程相同的JMX端口.出现端口已在使用中"错误
This error could be caused by any reason that prevents the Spark interpreter process from starting.Recently, I ve got it when trying to add the JMX options to ZEPPELIN_JAVA_OPTS, that cause the interpreter process to use the same JMX port as the Zeppelin process. Giving the "Port Already in Use" error
请检查Zeppelin日志(默认情况下，它们位于ZEPPELIN_DIR/logs/中，以查看Zeppelin尝试启动Spark Interpreter时发生的情况
Please check the Zeppelin logs (by default they are in ZEPPELIN_DIR/logs/ to see what is happening when Zeppelin trying to start Spark Interpreter
这篇关于齐柏林飞艇的世界您好！的文章就介绍到这了，希望我们推荐的答案对大家有所帮助，也希望大家多多支持！