The distributed deployment of hadoop is on three machines: hadoop.node.server01, hadoop.node.server02, hadoop.node.server03, the nameNode is deployed in server01, and the resourceManager of YARN is deployed in server02 and configured on the server01 machine Good configuration files such as core-site.xml, hdfs-site.xml, mapred-site.xml, slave, etc.
Configure passwordless login, scp the configuration file to other machines, and execute bin/hdfs namenode -format
Quickly tap the start commands sbin/start-dfs.sh and sbin/start-yarn.sh on the server01 machine, and then use the jps command on each machine to check whether the service is normal. It turns out that the resourceManager is not displayed. I immediately realized whether there was a problem with the configuration file. I checked the configuration items, host name, etc. of all the previous configuration files. No problem was found. Later, I tried to start each machine one by one, and found that the resourceManager was normal in server02. Only then did I understand that the previous batch command such as start-yarn.sh cannot be started on the machine where the namenode is located, but must be It is normal to type this command in server02 where the resourceManager is located.