Wednesday, 27 June 2012

Hadoop Hangover: Introduction To Apache Bigtop and Installing Hive, HBase and Pig

In the previous post we learnt how easy it was to install Hadoop with Apache Bigtop!
We know its not just Hadoop and there are sub-projects around the table! So, lets have a look at how to install Hive, Hbase and Pig in this post.

Before rowing your boat...
Please follow the previous post and get ready with Hadoop installed!
Follow the link for previous post:
also, the same can be found at DZone, developer site:

All Set?? Great! Head On..
Make sure all the services of Hadoop are running. Namely, JobTracker, SecondaryNameNode, TaskTracker, DataNode and NameNode. [standalone mode]

Hive with Bigtop:
The steps here are almost the same as Installing Hive as a separate project.
However, few steps are reduced.
The Hadoop installed in the previous post is Release 1.0.1

We had installed Hadoop with the following command
sudo apt-get install hadoop\*
Step 1: Installing Hive
We have installed Bigtop 0.3.0, and so issuing the following command installs all the hive components.
ie. hive, hive-metastore, hive-server. The daemons names are different in Bigtop 0.3.0.
sudo apt-get install hive\*
This installs all the hive components. After installing, the scripts must be able to create /tmp and /usr/hive/warehouseand HDFS doesn't allow these to be created while installing as it is unaware of the path to Java. So, create the directories if not created and grant the execute permissions.
In the hadoop directory, ie. /usr/lib/hadoop/
bin/hadoop fs -mkdir /tmp
bin/hadoop fs -mkdir /user/hive/warehouse

bin/hadoop -chmod g+x /tmp
bin/hadoop -chmod g+x /user/hive/warehouse

Step 2: The alternative directories could be/var/run/hiveand/var/lock/subsys
sudo mkdir /var/run/hive
sudo mkdir /var/lock/subsys

Step 3: Start the hive server, a daemon
sudo /etc/init.d/hive-server start
start hive-server

Step 4: Running Hive
Go-to the directory /usr/lib/hive.
See the Image below:

Step 5: Operations on Hive
Basic hive operations

HBase with Bigtop:
Installing Hbase is similar to Hive.

Step 1: Installing HBase
sudo apt-get install hbase\*

Step 2: Starting HMaster
sudo service hbase-master start
Starting HMaster

jps (HMaster started)

Step 3: Starting HBase shell
hbase shell
start HBase shell

Step 4: HBase Operations
HBase table operations

list,scan,get,describe In HBase

Pig with Bigtop:
Installing Pig is similar too.

Step 1: Installing Pig
sudo apt-get install pig
Installing Pig

Step 2: Moving a file to HDFS
Moving a tab separated file "book.csv" to HDFS

Step 3: Installed Pig-0.9.2
Pig installed Pig-0.9.2

Step 4: Starting the grunt shell
Starting Pig

Step 5: Pig Basic Operations
Basic Pig Operations

Job Completion

We saw that is it possible to install the subprojects and work with Hadoop, with no issues.
Apache Bigtop has its own spark! :)
There is a release coming BIGTOP-0.4.0 which is supposedly to fix the following issues:
Source and binary files:
Maven staging repo:
Bigtop's KEYS file containing PGP keys we use to sign the release:

Let us see how to install other sub-projects in the coming posts!
Until then, Happy Learning!! :):)

No comments:

Post a Comment