See John's comment below. On Mon, Apr 11, 2011 at 11:14 AM, John Sichi <jsi...@fb.com> wrote:
> Until HBase has a well-defined separation between client and server, > including protocol compatibility across versions, the situation is going to > remain sticky. > > I think I heard that 0.89 and 0.90 should be protocol compatible, but I > haven't confirmed that. If it's true, then you should be able to just use > Hive 0.7 as is (with the 0.89 jars) against an HBase 0.90 cluster. > > If that's not true, then follow the procedure described in the wiki page to > rebuild Hive from source after editing ivy/libraries.properties to change > the hbase.version property. > > I'm not sure about the Hadoop append part; maybe someone else knows the > answer regarding the Hive/HBase dependencies there. > > JVS > > On Apr 10, 2011, at 9:49 AM, <hi...@gmx.de> > wrote: > > > Hello, > > > > I am using the hive/hbase integration (which is a great feature) on a > standalone one machine environment. For production I set up a hadoop cluster > and now hbase. Here > > http://wiki.apache.org/hadoop/Hive/HBaseIntegration > > I read I should use hbase 0.89. > > And here > > http://hbase.apache.org/book/notsoquick.html > > its written you must use "the sync-supporting Hadoop jar". > > Does the following setup will work? > > I chose hbase 0.90 (I couldn't find 0.89). Because of the sync-supporting > I would take the hadoop-core-0.20-append.jar from hbase 0.90 release and > replace the hadoop jars in my hadoop installations with that. The hbase 0.90 > jar I will replace with the 0.89 jar from the hive 0.7.0 release. Will it > come to compatibility trouble with this version setup? And what about > stability and version updates in future... > > > > Thanks > > labtrax > > > > -- > > Empfehlen Sie GMX DSL Ihren Freunden und Bekannten und wir > > belohnen Sie mit bis zu 50,- Euro! https://freundschaftswerbung.gmx.de > >
