Compile

mvn package

Note: The following two jars may not be found in the maven repository. You need to modify pom.xml to solve the dependency problem during compilation, you may refer to the comment in pom.xml.

1. libchubaofs-xx-SNAPSHOT.jar, building from chubaofs/java
2. hadoop-common-3.0.0-tests.jar, from hadoop bin package(hadoop-3.3.0/share/hadoop/common/hadoop-common-3.3.0-tests.jar)

Deploy

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

One so:

1. libsdk.so, building from chubaofs/libsdk, used in cfs-site.xml.

Two jars:

1. chubaofs-hadoop-xx-SNAPSHOT.jar
2. libchubaofs-xx-SNAPSHOT.jar

Two profiles:

1. core-site.xml
2. cfs-site.xml

HDFS Shell on ChubaoFS

1. Put the two jars to $HADOOP_HOME/share/hadoop/common
2. Put the two profiles to $HADOOP_HOME/etc/hadoop

YARN on ChubaoFS

1. Put the two jars to $HADOOP_HOME/share/hadoop/common
2. Put the two profiles to $HADOOP_HOME/etc/hadoop
3. Start the yarn servers.

HBase on ChubaoFS

1. Put the two jars to $HBASE_HOME/lib.
2. Put the two profiles to $HBASE_HOME/conf.
3. Modify conf/hbase-site.xml `` hbase.rootdir` `cfs://172.16.1.1:8080/hbase` `
4. Start hbase。

Usage

Just replace hdfs:// with cfs://

About tests.

1. The unit tests.

   The scirpt is sbin/unit_test

2. Mapreduce Benchmark

The following cases are passed: dfsio/mrbench/nnbench/terasort You may view detailed test parameters in the sbin/mr_bench

3. HBase Benchmark

The following cases are passed: randomWrite/sequentialWrite/randomRead/sequentialRead/checkAndDelete You may view detailed test parameters in the sbin/hbase_bench

Note: The above tests are done in a cluster of 4 dockers, each uses 4 cpus and 8GB memory.