> I ended up creating a pseudo-distributed installation on Ubuntu in a > Virtual Box. It all works fine from localhost, and I can run the shell. But I > don't see how that's useful to anyone who actually wants to build a real > application. I'm struggling to > figure out how to "connect" to it from a remote Java client. The > documentation in this area seems to be sorely lacking, and the painfully > brief "Also note:" comment in the quick start doesn't seem to actually work. > (preemptive comment: yes, I know Virtual Box and can connect to the Ubuntu > machine in every other way I've tried)
I haven't tried in a while but as far as I know you must run a fully-distributed operation to be able to connect from "outside". After that the steps are "fairly" easy. You have to include the hbase jar along with all its dependencies (I know those aren't documented anywhere right now: zookeeper, hadoop-core, hadoop-hdfs, hadoop-mapred(?), commons-logging, commons-logging-api, log4j) and you need a hbase-site.xml in your classpath that includes the property " hbase.zookeeper.quorum" which needs to point to your zookeeper quorum (in a single node testcase this would be only a single name/ip address). I'm sure the XML file isn't necessary and a Configuration object can be build manually... After that it should work. I'm developing mainly on Windows too and it works just fine. Building HBase etc. works too, only the tests won't run. There are improvements planned to refactor out a client jar that can be easily used. Feel free to contact me if you need further help with accessing HBase from Windows. Cheers, Lars
