Am I correct in assuming that disabling the Hbase sink unit tests will fix the issue? That is, even if the Hbase artifacts used in the build is built against hadoop-1, it would work fine against hbase built against hadoop-2 (since hbase did not change any interfaces to abstract classes, hadoop did right?).
Thanks, Hari -- Hari Shreedharan On Tuesday, October 16, 2012 at 4:56 PM, Hari Shreedharan wrote: > Hi Roman, > > Even we would like to make sure we can build against hadoop-2. Do you have > suggestions on what we can do to fix this so that you can have flume-1.3.0 in > the next Bigtop release? > > > Thanks, > Hari > > -- > Hari Shreedharan > > > On Tuesday, October 16, 2012 at 4:16 PM, Roman Shaposhnik wrote: > > > Guys, > > > > is there any chance that the issue with hadoop-0.23 > > profile being broken can be fixed in 1.3.0? I'm planning > > a Bigtop 0.5.0 release (due in a couple of month) and > > it would be very unfortunate if we have to stick with > > Flume 1.1.0. > > > > Please let me know if you need any help is solving this. > > > > Please also note, that this is simply an issue of proper > > names in POMs not lack of HBase artifacts. To Bigtop > > (or anybody else for that matter) the lack of HBase > > artifacts compiled against hadoop 2 is simple matter > > of actually building HBase before we build Flume > > and thus guaranteeing local copies of things that are > > correct in our .m2. > > > > Thanks, > > Roman. > > > > > > > >
