工人无法连接到主火花独立集群 [英] workers can't connect to master in spark standalone cluster
问题描述
我安装了Apache 1.6的火花在2台Ubuntu的(IP 192.168.217.136主机名=工人的虚拟机)是matser和(IP = 192.168.217.139主机名=从站)工作者
我配置SSH和配置的文件夹奴隶和spark-env.sh
奴隶cofiguration是
i installed apache spark 1.6 on 2 machines ubuntu (ip 192.168.217.136 hostname= worker-virtual-machine ) is the matser and (ip= 192.168.217.139 hostname =slave) is worker i configured ssh and configured folders slaves and spark-env.sh the cofiguration of slaves is
192.168.217.139
并为spark-env.sh文件夹
192.168.217.139 and for the spark-env.sh folder
出口SPARK_MASTER_IP = 192.168.217.136
export SPARK_MASTER_IP=192.168.217.136
出口SPARK_WORKER_CORES = 1
export SPARK_WORKER_CORES=1
出口SPARK_WORKER_MEMORY =800米
export SPARK_WORKER_MEMORY=800m
出口SPARK_WORKER_INSTANCES = 2
export SPARK_WORKER_INSTANCES=2
当我开始通过./sbin/start-all.sh集群并输入Web GUI,我发现只有1工人的连接,以便与192.168.217.139工人ipadress不到风度大师ipadress连接到主<一个HREF =http://i.stack.imgur.com/1GIAe.png相对=nofollow>在这里输入的形象描述
when i start the cluster by ./sbin/start-all.sh and enter the web GUI i found just 1 worker with ipadress of master connected so the worker with 192.168.217.139 ipadress dosn't connect to the master enter image description here
我试图./start-master.sh开始掌握并开始从工人机器工人
./start-slaves.sh火花://192.168.217.136:7077
并输入时JPS命令它给我的工人在终端创建但是当我移动到主终端,做这JPS告诉我只是掌握和图形用户界面不显示任何东西,当我回到机器的工人,做JPS我不找工人
i tried to start master by ./start-master.sh and start worker from worker machine by ./start-slaves.sh spark://192.168.217.136:7077 and when enter jps command it show me worker created in terminal but when i move to the terminal of master and do jps it show me just master and the GUI doesn't show anything when i return to worker machine and do jps i don't find the worker
问题出在哪里?
当我从它的机器启动工作manualy我在记录该信息错误
when i start worker manualy from its machine i have this message error in logs
推荐答案
的文档,你可以看到, ./ start-slave.sh
的ARGS取主IP地址
作为第一个参数。这意味着你从机器上运行./start-slave.sh并将其指向主计算机。它看起来像你做相反:在主机上运行,并指着从
Reading the docs, you can see that the args for ./start-slave.sh
takes the master IP address
as the first argument. This means you run ./start-slave.sh on the slave machine and point it to the master machine. It looks like you're doing the opposite: running on the master and pointing to the slave.
这篇关于工人无法连接到主火花独立集群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IT屋!