This is a preliminary evaluation by the Big Data SIG of the work required to get Apache HBase into Fedora. This work has been done based upon the 0.96 branch building against the hadoop 2.0.5-alpha package in Fedora 20.
How to Build
This evaluation was done using mvn-local building against the 0.96 branch from the Big Data SIG HBase git repository. To build, execute:
mvn-local -Dhadoop.profile=2.0 clean install
Add -DskipServerTests to skip the hbase-server test suite. Some of these tests fail and if you want to evaluate the integration tests you'll need to skip them.
Issues To Be Resolved
Missing Java Dependencies
|htrace, htrace-core, htrace-zipkin||Complete||RHBZ #1016767||rrati|
|jruby-common, jruby-complete||Complete||rrati||jruby 1.7.2 is packaged in Fedora, but 1.6.8 is requested by Hbase. gid:aids differ, might need porting. Ported to later jruby|
|libthrift||Complete||RHBZ #982285||willb||Will Benton has some RPM artifacts at http://freevariable.com/thrift/|
Dependency Version Mismatches
HBase wants to use commons-math2. It will need to be ported to the latest version of commons-math in Fedora.
Need to correct import statements for commons-logging. Should be compile issue only.
HBase wants to use jetty 6. It will need to be ported to support the latest version of jetty in Fedora.
HBase wants to use tomcat/jasper version 5.x. It will need to be ported to use the latest version of tomcat in Fedora.
Numerous test failures pop out when running the test suite against local rpms. These will need to be investigated and compared against the baseline build.