On Fri, 6 May 2016, Adam C. Emerson wrote: > On 06/05/2016, Sage Weil wrote: > [snip] > > and we'll also need to keep the v1 protocol support around for a > > long time (probably a couple years at least) for the benefit of old > > clients. And there's no reason anybody *has* to switch to the new > > protocol unless they need the new features or improved security. > [snip] > > I can understand not wanting to force people to take a performance > hit. However, it sounded, from your writeup, like even without > encryption the new auth mechanism would be an improvement on the old > one. Once we do get it working, tested, and performant, I'd suggest we > might want to force users' hand on the issue and announce that the old > mechanism will be removed in some upcoming version, rather than > waiting for users to rotate older clients out over time. As soon as the new stuff is stable and performant, yeah, I think we should announce the old stuff is deprecated. But we will have to keep it around for a pretty long time to support existing kernels and clients. People still pretty frequently have dumpling or older clients talking to up-to-date clusters. Partly this is just infrastructure moving slowly, and partly it users not wanting to restart running VMs. I would say at least 2 LTS releases, maybe longer... sage -- To unsubscribe from this list: send the line "unsubscribe ceph-devel" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
