I know nothing about etcd/raft either. My motivation is more future thinking.
> It would seem like a pretty large amount of work (essentially > a complete rewrite) I think the recipes can be saved. Very little of ZooKeeper is exposed in most of them (other than the paths). Internally, the Curator Framework stuff could be abstracted. I don’t think it would be that bad. > I wonder if we've collectively got enough spare cycles There’s no rush. It could be done in the background. It might be just you and me unless someone else steps up. I’m in for it if you are. -Jordan > On Jul 27, 2016, at 6:15 PM, Cameron McKenzie <[email protected]> wrote: > > hey Jordan, > Sounds like a good idea in theory (though I have no experience in > etcd/consol), but I wonder if we've collectively got enough spare cycles to > get it done. It would seem like a pretty large amount of work (essentially > a complete rewrite) to abstract all of the ZK concepts into something that > is shared across different backends. > cheers > > On Thu, Jul 28, 2016 at 3:15 AM, Jordan Zimmerman < > [email protected]> wrote: > >> Hello Devs, >> >> etcd/consul is starting to gain traction. Currently, there is no Java >> driver for it. Even if they do make a driver I doubt they’ll have high >> level patterns for it. I’m starting to wonder if we could refactor Curator >> so that it has a pluggable backend such that the same code (or close >> variations) could run on either ZooKeeper or etcd or whatever. >> >> Thoughts? >> >> -Jordan
