Installation of CarbonData 1.1.0 with Spark 1.6.2

关键词：carbondata spark thrift 数据仓库

【Install thrift 0.9.3】

注意 要装thrift-java必须先装ant 。

有人说要装boost，我在centos6上没有装一样可以运行，猜测可能是c/cpp需要，java/python的不需要

thrift安装包可以在thrift官网下载，注意版本，手动下载地址：http://www.apache.org/dyn/closer.cgi?path=/thrift/0.9.3。

sudo yum -y install ant libevent-devel zlib-devel openssl-devel

# Install bison
wget http://ftp.gnu.org/gnu/bison/bison-2.5.1.tar.gz
tar xvf bison-2.5.1.tar.gz
cd bison-2.5.1
./configure --prefix=/usr
make
sudo make install
cd ..

# Install libevent
wget --no-check-certificate https://github.com/libevent/libevent/releases/download/release-2.0.22-stable/libevent-2.0.22-stable.tar.gz -O libevent-2.0.22-stable.tar.gz
tar -xzvf libevent-2.0.22-stable.tar.gz
cd libevent-2.0.22-stable
./configure --prefix=/usr
make
sudo make install
cd ..

# Install thrift
wget http://apache.parentingamerica.com/thrift/0.9.3/thrift-0.9.3.tar.gz
tar -xzvf thrift-0.9.3.tar.gz
cd thrift-0.9.3
./configure --prefix=/usr --with-libevent=/usr --with-java
sudo make
sudo make install
cd ..

如果是其他语言的，首先得安装该语言的环境和其他相关的库。Java的需要jdk和ant。

【Package and Install CarbonData】

参考：https://github.com/apache/carbondata/tree/master/build

下载 carbondata 1.1.0，解压后在carbondata源码目录下执行 (同理其他spark版本改下profile和spark.version的参数即可)

mvn -DskipTests -Pspark-1.6 -Dspark.version=1.6.2 clean package

maven下载速度慢的，可以用aliyun mirror替代apache central，修改 ~/.m2/settings.xml。

<settings>
  ...
  <mirrors>
    <mirror>
      <id>alimaven</id>
      <name>aliyun maven</name>
      <url>http://maven.aliyun.com/nexus/content/groups/public/</url>
      <mirrorOf>central</mirrorOf>
    </mirror>
  </mirrors>
  ...
</settings>

【Run carbondata in spark-shell】

参考：http://carbondata.apache.org/quick-start-guide.html

准备数据文件

# in linux , prepare data example file
cd carbondata
cat > sample.csv << EOF
id,name,city,age
1,david,shenzhen,31
2,eason,shenzhen,27
3,jarry,wuhan,35
EOF

hdfs dfs -put sample.csv /tmp/

准备assembly jar包

# in linux, copy assembly jar to a lib directory
cd $CARBONDATA_HOME
mkdir -p lib
cp assembly/target/scala-2.10/carbondata_2.10-1.1.0-shade-hadoop2.2.0.jar lib/
cp integration/spark/target/carbondata-spark-1.1.0.jar lib/

run spark in shell mode

spark-shell --jars $CARBONDATA_HOME/lib/carbondata_2.10-1.1.0-shade-hadoop2.2.0.jar,$CARBONDATA_HOME/lib/carbondata-spark-1.1.0.jar

SparkShell > 

// in spark shell, cluster mode
import org.apache.spark.sql.CarbonContext

// remember to add hdfs:// if you want to use hdfs mode.
val cc = new CarbonContext(sc, "hdfs:///tmp/carbon/data/")
cc.sql("CREATE TABLE IF NOT EXISTS hdfs_sample ( id string, name string, city string, age Int) STORED BY 'carbondata'")
cc.sql("LOAD DATA INPATH 'hdfs:///tmp/sample.csv' INTO TABLE hdfs_sample")
cc.sql("SELECT * FROM hdfs_sample").show()
cc.sql("SELECT city, avg(age), sum(age) FROM hdfs_sample GROUP BY city").show()