Configuring Hadoop High Availability Cluster(Two Namenode service)

Configuring High Availability Hadoop Cluster:(Qjournal Method)

Steps:
Step1: Download and configure Zookeeper
Step2: Hadoop configuration and high availability settings
Step3: Creating folders for Hadoop cluster and file permissions
Step4: Hdfs service and file system format

Steps in Details:
Step 1: Download and configure Zookeeper

1.1 Download and configure Zookeeper software package from (https://www.apache.org/dist/zookeeper/zookeeper-3.4.5/zookeeper-3.4.5.tar.gz)
[cluster@n3:~]$wget https://www.apache.org/dist/zookeeper/zookeeper-3.4.5/zookeeper-3.4.5.tar.gz
Extract source
[cluster@n3:~]$tar –zxvf zookeeper-3.4.5.tar.gz

1.2 Zookeeper related configuration files are located
Configuration files: /home/cluster/zookeeper-3.4.5/conf
Binary executables: /home/cluster/zookeeper-3.4.5/bin
The Main configuration file
/home/cluster/zookeeper-3.4.5/conf/zoo.cfg
[cluster@n3:~]$cp zoo_sample.cfg zoo.cfg
Modifying zoo.cfg as per our installation guide
[cluster@n3:~]$nano /home/cluster/zookeeper-3.4.5/conf/zoo.cfg
tickTime=2000
clientPort=3000
initLimit=5
syncLimit=2
dataDir=/home/cluster/zookeeper/data/
dataLogDir=/home/cluster/zookeeper/log/
server.1=n3:2888:3888
server.2=n4:2889:3889
Save & Exit!
Note :-If Each of the servers hosted in the same physical machine as instance , every server port number has changed to n3:2888:3888 , n4: 2889:3889


1.3 Create the folder structure for Zookeeper data and logs as defined in zoo.cfg , repeat following step in all the nodes in the cluster (n3 & n4)
[cluster@n3:~]$mkdir –p /home/cluster/zookeeper/data/
[cluster@n3:~]$mkdir –p /home/cluster/zookeeper/log/

1.4 Create the myid file in /home/cluster/zookeeper/data/ and assign the value of each of the nodes in cluster. (n3=1 & n4=2)
[cluster@n3:~]$nano /home/cluster/zookeeper/data/myid
1
Save and Exit!
[cluster@n4~]$nano /home/cluster/zookeeper/data/myid
2
Save & Exit!

Step 2: Hadoop configuration and high availability settings

Download the latest hadoop 2.X.X version from the releases
2.1 Add / modify the following lines in hadoop-env.sh file to apply environment variable settings.
[cluster@n3:~]$ nano /home/cluster/hadoop-2.2.0-cdh5.0.0-beta-2/etc/hadoop/hadoop-env.sh
export JAVA_HOME=/usr/lib/jvm/jdk1.7.0_45/
export HADOOP_COMMON_LIB_NATIVE_DIR=/home/cluster/hadoop-2.2.0-cdh5.0.0-beta-2/lib/native/
export HADOOP_OPTS=”-Djava.library.path=/home/cluster/hadoop-2.2.0-cdh5.0.0-beta-2/lib/native/”
2.2 Add following lines in cores-site.xml file to configure journaling, default FS, temp directory & hdfs cluster. Within the <configuration> tag
[cluster@n3:~]$nano /home/cluster/hadoop-2.2.0-cdh5.0.0-beta-2/etc/hadoop/core-site.xml
<property>
    <name>hadoop.tmp.dir</name>
    <value>/hdfs/dfs/tmp</value>
</property>
<property>
    <name>fs.defaultFS</name>
    <value>hdfs://mycluster</value>
</property>
<property>
  <name>dfs.journalnode.edits.dir</name>
  <value>/hdfs/dfs/journal/data</value>
</property>
2.3 Add following lines in hdfs-site.xml file to configure dfs nameservice, dfs high availability, zookeeper & failover. Within the <configuration> tag.
[cluster@n3:~]$nano /home/cluster/hadoop-2.2.0-cdh5.0.0-beta-2/etc/hadoop/hdfs-site.xml
<property>
    <name>dfs.name.dir</name>
    <value>/hdfs/dfs/nn</value>
</property>
<property>
    <name>dfs.data.dir</name>
    <value>/hdfs/dfs/dn</value>
 </property>
 <property>
    <name>dfs.nameservices</name>
    <value>mycluster</value>
    <final>true</final>
 </property>
<property>
  <name>dfs.ha.namenodes.mycluster</name>
  <value>n3,n4</value>
  <final>true</final>
</property>
 <property>
    <name>dfs.namenode.rpc-address.mycluster.n3</name>
    <value>n3:8020</value>
 </property>
 <property>
    <name>dfs.namenode.http-address.mycluster.n3</name>
    <value>n3:50070</value>
 </property>
 <property>
    <name>dfs.namenode.secondaryhttp-address.mycluster.n3</name>
    <value>n3:50090</value>
 </property>
 <property>
    <name>dfs.namenode.rpc-address.mycluster.n4</name>
    <value>n4:8020</value>
 </property>
 <property>
    <name>dfs.namenode.http-address.mycluster.n4</name>
    <value>n4:50070</value>
 </property>
 <property>
    <name>dfs.namenode.secondaryhttp-address.mycluster.n4</name>
    <value>n4:50090</value>
 </property>
<property>
  <name>dfs.namenode.shared.edits.dir</name>
  <value>qjournal://n3:8485;n4:8485/mycluster</value>
</property>
<property>
  <name>dfs.ha.automatic-failover.enabled</name>
  <value>true</value>
</property>
<property>
  <name>ha.zookeeper.quorum</name>
  <value>n3:3000,n4:3000</value>
</property>
<property>
  <name>dfs.ha.fencing.methods</name>
  <value>sshfence</value>
</property>
<property>
  <name>dfs.ha.fencing.ssh.private-key-files</name>
  <value>/home/cluster/.ssh/id_rsa</value>
</property>
<property>
  <name>dfs.ha.fencing.ssh.connect-timeout</name>
  <value>30000</value>
</property>
<property>
<name>dfs.client.failover.proxy.provider.mycluster</name>
<value>org.apache.hadoop.hdfs.server.namenode.ha.ConfiguredFailoverProxyProvider</value>
</property>

2.4 Add datanodes in the slaves configuration file as shown below.
[cluster@n3:~]$nano /home/cluster/hadoop-2.2.0-cdh5.0.0-beta-2/etc/hadoop/slaves
n3
n4
Save & Exit!

Step 3 : Creating folders for Hadoop cluster and set file permissions

3.1 Create folder structure for journalnode as defined in core-site.xml, repeat following step in all the cluster nodes (n3 & n4)
[cluster@n3:~]$mkdir –p /hdfs/dfs/journal/data

3.2 Create temp folder for hadoop cluster as defined in core-site.xml, repeat following step in all the cluster nodes (n3 & n4)
[cluster@n3:~]$mkdir -p /hdfs/dfs/tmp

3.3 Create datanode and namenode folder for hadoop cluster as defined in hdfs-site.xml, repeat following step in all the cluster nodes (n3 & n4)
[cluster@n3:~]$mkdir -p /hdfs/dfs/dn
[cluster@n3:~]$mkdir -p /hdfs/dfs/nn

3.4 Copy hadoop source and zookeeper source configured in n3 node to n4

Step 4: Hdfs service and file system format

4.1 Start zookeeper service, once in all the nodes in cluster used for zookeeper, repeat below step in all the cluster nodes running zookeeper (n3 & n4).
Go to zookepeer-3.4.5 Binary path i.e /home/cluster/zookeeper-3.4.5/bin, then execute below commands
[cluster@n3:~]$./zkServer.sh start
[cluster@n4~]$./zkServer.sh start

4.2 Format Zookeeper file system in n3
Go to hadoop home path i.e /home/cluster/hadoop-2.2.0-cdh5.0.0-beta-2, then execute below command
[cluster@n3:~]$bin/hdfs zkfc –formatZK
Before format start journalnode in all the cluster nodes (n3 & n4)
[cluster@n3:~]$sbin/hadoop-daemon.sh start journalnode

4.3 Format namenode in n3
Go to hadoop home path i.e /home/cluster/hadoop-2.2.0-cdh5.0.0-beta-2, then execute below command
[cluster@n3:~]$bin/hdfs namenode –format

4.4 Copy Meta data information to slave name node in our guide (n4), run below command in
n4 (slave).
Make sure that namenode service is running in master node(n3)….
Go to hadoop home path i.e /home/cluster/hadoop-2.2.0-cdh5.0.0-beta-2, then execute below command
[cluster@n3:~]$sbin/hadoop-daemon.sh start namenode
Then in n4,
Go to hadoop home path i.e /home/cluster/hadoop-2.2.0-cdh5.0.0-beta-2, then execute below command
[cluster@n4~]$bin/hdfs namenode -bootstrapStandby


Start hadoop service............


[cluster@n3:~]$cd /home/cluster/hadoop-2.2.0-cdh5.0.0-beta-2/sbin
./stop-all.sh
and start again.
./start-dfs.sh
run jps to check services running in n3 & n4
[cluster@n3 sbin]$ jps
17588 DFSZKFailoverController
16966 DataNode
24142 Jps
4268 QuorumPeerMain
16745 NameNode
17276 JournalNode
[cluster@n4 bin]$ jps
14357 DFSZKFailoverController
2369 QuorumPeerMain
13906 DataNode
23689 Jps
15458 NameNode
14112 JournalNode



Verifying Automatic Failover

After the initial deployment of a cluster with automatic failover enabled, you should test its operation. To do so, first locate the active NameNode. As mentioned above, you can tell which node is active by visiting the NameNode web interfaces.
Once you have located your active NameNode, you can cause a failure on that node. For example, you can use kill -9 <pid of NN> to simulate a JVM crash. Or you can power-cycle the machine or its network interface to simulate different kinds of outages. After you trigger the outage you want to test, the other NameNode should automatically become active within several seconds. The amount of time required to detect a failure and trigger a failover depends on the configuration of ha.zookeeper.session-timeout.ms, but defaults to 5 seconds.
If the test does not succeed, you may have a misconfiguration. Check the logs for the zkfc daemons as well as the NameNode daemons in order to further diagnose the issue.




Thank You.