Note: My public key file is not under/root/hxsyl under/home/hxsyl/.ssh,
Find/-name Id_rsa
Find
1.
Run the command on each machine zkserver.sh start or run in the $zookeeper_home/bin directory./zkserver.sh start command. You can then use the command JPS to view zookeeper-initiated process Quorumpeermain.
The zookeeper status can be viewed through the zkserver.sh status command. Normally there is only one leader in the machine, and the others are follow.
2. Master node execution
HDFs Zkfc-formatzk
Note: The final ZK is capitalized, otherwise
6/11/30 20:31:45 FATAL Tools. Dfszkfailovercontroller:got a fatal error, exiting Noworg.apache.hadoop.HadoopIllegalArgumentException:Bad argument: -formatzk
He initializes it based on the value Ha.zookeeper.quorum in the $hadoop_home/etc/hadoop/core-site.xml file.
This needs to be sure that the automatic ha switch is turned on or not.
But now
Dfszkfailovercontroller does not start, start ZKFC only then can, then is an active, a standby,
3.
Start Journalnode
or in Mster execution
Start Journalnode
Note the second boot method the master node is not journalnode and needs to be started separately, it's worth the time to stop alone
4. on [nn1], format it, and start: bin/hdfs namenode-< Span class= "Hljs-keyword" >formatsbin/hadoop-daemon.sh start namenode5. On [NN2], Synchronize nn1 metadata information: bin/hdfs Namenode-bootstrapstandby 6. start [nn2]:sbin/hadoop-daemon.sh start Namenode after the above four steps, NN1 and nn2 are processed standby status 7. switch [NN1] to active
There's a problem with how to configure automatic switching, this does not, forcing manual brain problems ....
Bin/hdfs haadmin-transitiontoactive nn18. Start Datanode
9. Start yarn
sbin/start-yarn.sh
More ResourceManager process on Master1, slave3 process on slave1 slave2 NodeManager
10. Start ZKFC
sbin/hadoop-daemons.sh Start ZKFC
Note that this and the above boot Journalnode specifications, Damons does not start the master ZKFC, need to be started separately.
11.
sbin/mr-jobhistory-daemon.sh start Historyserver Original configuration on the stand, I changed to Centosmaster.
11. Start HBase
bin/start-hbase.sh
Close the Hadoop cluster: On [NN1], enter the following command sbin/stop-dfs.sh does not stop the journalnode of the above yarn and history Server master node
进入spark的sbin目录下
start-all.sh
./start-history-server.sh
Bin/spark-shell
12. Close
A.zookeeper
bin/zkserver.sh stop
Start Hadoop HA Hbase zookeeper Spark