cubefs-hadoop

Hadoop API package support mulit data layer such as blobstore (Erasure-Code) to adapte to datalake.

Compile

mvn package -Dmaven.test.skip=true

Notes: You are suggested to skip the test.

Deploy

When using the SDK, you need to use two jars and one so and one profile.

Dependent packages

cubefs-hadoop-x.x.x.jar(build by mvn pakcage)
libcfs.so(build from cubefs/libsdk)
jna-x.x.x.jar(you can find it in the maven repository, minimum version is 4.0)

Configuration

Modify the configuration file(core-site.xml or hdfs-site.xml) and add the following items to it.

<property>
	<name>fs.cfs.impl</name>
	<value>io.cubefs.CubefsFileSystem</value>
</property>

<property>
	<name>cfs.master.address</name>
	<value>your.master.address[ip:port,ip:port,ip:port]</value>
</property>

<property>
	<name>cfs.log.dir</name>   
	<value>your.log.dir[/tmp/cfs-access-log]</value>
</property>

<property>
	<name>cfs.log.level</name> 
	<value>INFO</value>
</property>

<property>
    <name>cfs.access.key</name>
    <value>your.access.key</value>
</property>

<property>
    <name>cfs.secret.key</name>
    <value>your.secret.key</value>
</property>

<property>
	<name>cfs.min.buffersize</name>
	<value>8388608</value>
</property>

HDFS Shell on CubeFS

Put the two jars to $HADOOP_HOME/share/hadoop/common/lib
Put the so to $HADOOP_HOME/lib/native
Modify the configuration file $HADOOP_HOME/etc/hadoop/core-site.xml

YARN on CubeFS

Same to HDFS Shell on CubeFS.

Spark on CubeFS

Put the three dependent packages to $SPARK_HOME/jars

Hive on CubeFS