com.bigstep:datalake-client-libraries

This is a support library for interacting with the Bigstep's DataLake from hadoop.

License

License

Categories

Categories

Data CLI User Interface
GroupId

GroupId

com.bigstep
ArtifactId

ArtifactId

datalake-client-libraries
Last Version

Last Version

1.4
Release Date

Release Date

Type

Type

jar
Description

Description

com.bigstep:datalake-client-libraries
This is a support library for interacting with the Bigstep's DataLake from hadoop.
Project URL

Project URL

https://github.com/bigstepinc/datalake-client-libraries
Source Code Management

Source Code Management

https://github.com/bigstepinc/datalake-client-libraries

Download datalake-client-libraries

How to add to project

<!-- https://jarcasting.com/artifacts/com.bigstep/datalake-client-libraries/ -->
<dependency>
    <groupId>com.bigstep</groupId>
    <artifactId>datalake-client-libraries</artifactId>
    <version>1.4</version>
</dependency>
// https://jarcasting.com/artifacts/com.bigstep/datalake-client-libraries/
implementation 'com.bigstep:datalake-client-libraries:1.4'
// https://jarcasting.com/artifacts/com.bigstep/datalake-client-libraries/
implementation ("com.bigstep:datalake-client-libraries:1.4")
'com.bigstep:datalake-client-libraries:jar:1.4'
<dependency org="com.bigstep" name="datalake-client-libraries" rev="1.4">
  <artifact name="datalake-client-libraries" type="jar" />
</dependency>
@Grapes(
@Grab(group='com.bigstep', module='datalake-client-libraries', version='1.4')
)
libraryDependencies += "com.bigstep" % "datalake-client-libraries" % "1.4"
[com.bigstep/datalake-client-libraries "1.4"]

Dependencies

compile (5)

Group / Artifact Type Version
commons-codec : commons-codec jar 1.10
org.apache.hadoop : hadoop-common jar 2.7.1
org.apache.hadoop : hadoop-hdfs jar 2.7.1
com.google.code.gson : gson jar 2.7
org.apache.directory.server : apacheds-kerberos-codec jar 2.0.0-M23

test (4)

Group / Artifact Type Version
org.apache.hadoop : hadoop-common test-jar 2.7.1
org.apache.spark : spark-core_2.10 jar 2.0.1
org.apache.spark : spark-sql_2.10 jar 2.0.1
junit : junit jar 4.12

Project Modules

There are no modules declared in this project.

Bigstep DataLake client libraries

These libraries enable the "dl://" prefix in hadoop and associated tools so that hdfs dfs -ls or distcopy work properly. They can also be used as a standalone FileSystem implementation to enable easy interaction with the DataLake from java or scala applications. They also build with a standalone version of "hdfs dfs".

Build Status

###Before getting started:

  1. Deploy a "DataLake" in the Bigstep's Control panel.

  2. Install Java Cryptography Extension on all nodes of the cluster.

###Using the standalone tool The command line tool should be independent of the environment. It can coexist with any kerberos instalation. To use: 0. Have java installed, have the unlimited strenght policy files installed (openJDK includes them)

  1. Download the binaries

latest SNAPSHOT

  1. Generate a keytab:
 ./bin/dl genkeytab [email protected] /etc/kxxxx.keytab
  1. Add your kerberos identity to the embeded core-site.xml:
vim conf/core-site.xml
  1. Execute any hdfs dfs command ./bin/dl fs -ls dl://node10930-datanodes-data-lake01-uk-reading.bigstep.io:14000/data_lake/dlxxx/

###Using as part of a Hadoop environment:

To use the library within a hadoop environment:

  1. Install kerberos client libraries: Centos:
yum install krb5-workstation

Ubuntu:

apt-get install krb5-config, krb5-user, krb5-clients

OsX:

brew install krb5
  1. Update the /etc/krb5.conf or download the auto-generated file from you Bigstep account:
[appdefaults]
validate=false

[libdefaults]
 default_realm =  bigstep.io
 dns_lookup_realm = false
 dns_lookup_kdc = false
 ticket_lifetime = 24h
 renew_lifetime = 7d
 forwardable = true
 validate=false
 rdns = false
 ignore_acceptor_hostname = true
 udp_preference_limit = 1
 
default_tgs_enctypes = aes256-cts-hmac-sha1-96 aes128-cts-hmac-sha1-96 des3-cbc-sha1 arcfour-hmac-md5 camellia256-cts-cmac camellia128-cts-cmac des-cbc-crc des-cbc-md5 des-cbc-md4
default_tkt_enctypes = aaes256-cts-hmac-sha1-96 aes128-cts-hmac-sha1-96 des3-cbc-sha1 arcfour-hmac-md5 camellia256-cts-cmac camellia128-cts-cmac des-cbc-crc des-cbc-md5 des-cbc-md4
permitted_enctypes = aes256-cts-hmac-sha1-96 aes128-cts-hmac-sha1-96 des3-cbc-sha1 arcfour-hmac-md5 camellia256-cts-cmac camellia128-cts-cmac des-cbc-crc des-cbc-md5 des-cbc-md4

[realms]
  bigstep.io = {
  kdc = fullmetal.bigstep.com
  admin_server = fullmetal.bigstep.com
  max_renewable_life = 5d 0h 0m 0s
 }
  1. Add the following to core-site.xml. If using Spark standalone place this file in the spark-2.x.x/conf directory:
<property>
  <name>fs.dl.impl</name>
  <value>com.bigstep.datalake.DLFileSystem</value>
</property>

<property>
  <name>fs.dl.impl.kerberosPrincipal</name>
  <value>[email protected]</value>
</property>

<property>
  <name>fs.dl.impl.kerberosKeytab</name>
  <value>/etc/hadoop/kx.keytab</value>
</property>

<property>
  <name>fs.dl.impl.kerberosRealm</name>
  <value>bigstep.io</value>
</property>

<property>
  <name>fs.dl.impl.homeDirectory</name>
  <value>/data_lake/dlxxxx</value>
</property>

<!-- optional -->
<property>
  <name>fs.dl.impl.defaultFilePermissions</name>
  <value>00640</value>
</property>

<!-- optional -->
<property>
  <name>fs.dl.impl.defaultUMask</name>
  <value>007</value>
</property>
  1. Create a keytab:
# ktutil
ktutil:  addent -password -p [email protected] -k 1 -e aes256-cts
ktutil:  wkt /root/.k5keytab
ktutil:  exit
  1. Make sure that the jar is available on all the cluster machines. Also the keytab must be reachable to yarn user (eg: not /root).
  • Vanilla Hadoop hadoop-2.7.x/share/hadoop/common/
  • Vanilla Spark 2.0 spark-2.x.x/jars
  • Cloudera CDH 5.x /opt/cloudera/parcels/CDH/lib/hadoop/
  • It should be possible technically to be used with any hadoop enabled application as long as it is added to the classpath.

You should now be able to use regular hadoop commands like distcp:

hadoop distcp hdfs://localhost/user/hdfs/test dl://node10930-datanodes-data-lake01-uk-reading.bigstep.io:14000/data_lake/dlzzz

##Using as part of a Spark environment:

  1. Extract a vanilla spark 2.* directory
  2. Copy the datalake-.jar in the jars directory of spark-2.
cp datalake-client-libraries-*.jar jars
  1. Modify core-site.xml to include the above
  2. Start spark-shell

###Using via Spark

./bin/spark-shell --packages com.bigstep:datalake-client-libraries:1.4

###Using programatically: To use directly in a maven use:

<dependency>
  <groupId>com.bigstep</groupId>
  <artifactId>datalake-client-libraries</artifactId>
  <version>1.4</version>
</dependency>

Javadoc is available from maven central.

To compile use:

mvn package

More information can be found at the DataLake documentation.

###Troubleshooting You can find out a lot more about what the library is doing by enabling debug:. Change the log4j.rootLogger to DEBUG in conf/log4j.properties file:

log4j.rootLogger=DEBUG, A1
Key for the principal [email protected] not available in /etc/kxxx.keytab
		[Krb5LoginModule] authentication failed
Unable to obtain password from user

Try enabling Kerberos debug via:

 export KERBEROS_DEBUG=true

If you see something along the lines of:

Found unsupported keytype (18) for [email protected]

You need to enable AES 256 (assuming it's not illegal in your country) by adding the "Unlimited Strength JCE" policy files. Follow your operating system guides for installing that. For Mac Try: http://bigdatazone.blogspot.ro/2014/01/mac-osx-where-to-put-unlimited-jce-java.html

If a command stalls for an unknown reason and then timeouts with KDC not found try adding the following in the [libdefaults] section of your /etc/krb5.conf file:

udp_preference_limit=1

If you see errors related to the missing StringUtil.toLowerCase try the hadoop2.6 branch as the master is linked to hadoop-2.7.

Using file encryption

It is possible to encrypt/decrypt files when uploading/downloading from the datalake. To enable this, add the following properties to core-site.xml:

  <!-- This tells the DataLake client if it should encrypt/decrypt files when
  uploading/downloading. If property is missing, the default value is false. -->
  <property>
    <name>fs.dl.impl.shouldUseEncryption</name>
    <value>true</value>
  </property>
  
  <!-- The location of the AES key. The file should be exactly 16 bytes long.
  This property is required if fs.fl.impl.shouldUseEncryption is set to true. -->
  <property>
    <name>fs.dl.impl.encryptionKeyPath</name>
    <value>/etc/PUT_YOUR_KEY_PATH_HERE</value>
  </property>

By setting fs.dl.impl.shouldUseEncryption to true, you should also provide the path to the file containing the AES key in fs.dl.impl.encryptionKeyPath (the file should be exactly 16 bytes long). This will enable encrypting or decrypting of files when uploading or appending, or, respectively, downloading them. You can easily disable this by setting fs.dl.impl.shouldUseEncryption to false.

com.bigstep

Bigstep Cloud

Big Data as a Service Company

Versions

Version
1.4
1.3
1.2
1.1