+1 from me, too, but I'd like to review what actually changes in master for the migration to happen. I don't know much about Hadoop 3. I'm curious what the releases will look like (AFAIK, it's only snapshot builds right now; is that correct?), how our dependencies will change, and what API stability guarantees it offers.
On Thu, Aug 3, 2017 at 12:59 PM Josh Elser <els...@apache.org> wrote: > +1 sounds like a good idea to me. > > On 8/3/17 10:08 AM, Sean Busbey wrote: > > Hi Folks! > > > > I think we need to start being more formal in planning for Hadoop 3. > > They're up to 3.0.0-alpha4 and are pushing towards API-locking betas[1]. > > > > What do folks think about starting to push on an Accumulo 2.0 release > that > > only supports Hadoop 3? It would let us move faster, which we'll need to > do > > if we want to get any API changes in to the Hadoop 3 line. > > > > If we get started soon we can probably make it parity on beta/GA status > > with the Hadoop 3 line. That would give us a beta for Accumulo Summit and > > GA by end of the year. > > > > Going to just Hadoop 3+ would also be a sufficient dependency break that > we > > could do a pass updating any lagging dependencies to latest major > version. > > > > > > [1]: > https://cwiki.apache.org/confluence/display/HADOOP/Hadoop+3.0.0+release > > >