This library allow to connect to the Hadoop datalab cluster without any system installation (except Java).
- A maven dependency can be import in your Java application for use hdfs
- A command line interface can be used on your machine hdfs dfs
<dependency>
<groupId>com.tony.hdfs</groupId>
<artifactId>HdfsClient</artifactId>
<version>1.0</version>
</dependency>
Define your hadoop.properties in your project
hadoop.cluster=clustername
hadoop.failoverProxy=org.apache.hadoop.hdfs.server.namenode.ha.ConfiguredFailoverProxyProvider
hadoop.namenodes=nn1,nn2
hadoop.rpcAddress=[DNS_NAMENODE1]:[PORT_RPC],[DNS_NAMENODE2]:[PORT_RPC]
hadoop.httpAddress=[DNS_NAMENODE1]:[PORT_HTTP],[DNS_NAMENODE2]:[PORT_HTTP]
hadoop.krb5Url=hadoop/krb5.conf
hadoop.jaasConfUrl=hadoop/jaas.conf
Note: Example krb5.conf and jaas.conf are embedded in jar and must be overriden
Properties prop = new Properties();
ClassLoader classLoader = getClass().getClassLoader();
InputStream input = new FileInputStream("./hadoop.properties");
prop.load(input);
client = new HadoopClient();
client.setHadoopCluster(prop.getProperty("hadoop.cluster"));
client.setNamenodes(prop.getProperty("hadoop.namenodes"));
client.setHttpAaddress(prop.getProperty("hadoop.httpAddress"));
client.setRpcAddress(prop.getProperty("hadoop.rpcAddress"));
client.setHadoopProxy(prop.getProperty("hadoop.failoverProxy"));
# For use internal krb5 and jaas files
URL jaas = classLoader.getResource(prop.getProperty("hadoop.jaasConfUrl"));
URL krb5 = classLoader.getResource(prop.getProperty("hadoop.krb5Url"));
# For use external krb5 and jaas files
#URL jaas = new File(prop.getProperty("hadoop.jaasConfUrl")).toURL();
#URL krb5 = new File(prop.getProperty("hadoop.krb5Url")).toURL();
client.setJaasConfUrl(jaas);
client.setKrbConfUrl(krb5);
String keytabPath = new File("xxx.keytab").getPath();
FileSystem fs = client.hadoopConnectionWithKeytab(keytabPath, "[email protected]");
// or with user/password
FileSystem fs = client.hadoopConnectionWithUserPassword("[email protected]", "xxx");
The project provide a fat jar with the original Hadoop client hdfs dfs interface usable on your local machine.
Copy next to the hadoop-client-cli.jar :
- your hadoop.properties
- the krb5.conf
- the jaas.conf
- your keytab xxx.keytab (if keytab authentication)
Example hadoop.properties for CLI with keytab
hadoop.cluster=clustername
hadoop.failoverProxy=org.apache.hadoop.hdfs.server.namenode.ha.ConfiguredFailoverProxyProvider
hadoop.namenodes=nn1,nn2
hadoop.rpcAddress=[DNS_NAMENODE1]:[PORT_RPC],[DNS_NAMENODE2]:[PORT_RPC]
hadoop.httpAddress=[DNS_NAMENODE1]:[PORT_HTTP],[DNS_NAMENODE2]:[PORT_HTTP]
hadoop.krb5Url=krb5.conf
hadoop.jaasConfUrl=jaas.conf
#Keytab auth
hadoop.keytab=xxx.keytab
[email protected]
#hadoop.password=XXX
Example hadoop.properties for CLI with user/pass authentication
#User/pass auth
#hadoop.keytab=xxx.keytab
[email protected]
hadoop.password=XXX
jaas.conf
HdfsHaSample {
com.sun.security.auth.module.Krb5LoginModule required client=TRUE debug=true;
};
java -jar hadoop-client-cli.jar -ls /
Deploy jar and files in %userprofile%/hdfs and add directory to Windows PATH.
Add hdfs.bat
@ECHO OFF
setlocal
cd /d %~dp0
java -jar %userprofile%/hdfs/hadoop-client-cli.jar %*
Usage in cmd:
hdfs -ls /