On Mon, Dec 29, 2014 at 3:24 PM, mck <m...@apache.org> wrote: > > Especially in CASSANDRA-6285 i see some scary stuff went down. > > But there are no outstanding bugs that we know of, are there? >
Right, the question is whether you believe that 6285 has actually been fully resolved. It's relatively plausible that it finally was, which is why I describe my feelings about HSHA "corrupter" implementation as FUD. Really the huge mistake was to rewrite "hsha" despite the fact that this is one of the rare pluggable interfaces, and thereby breaking existing users. If it had been called "hsha2" or something, I'd have a lot less FUD about it... because people would not have corrupted on upgrade, which I view as Super Bad. IMO, probably the only people who should use HSHA are people who have a real need for it, specifically people with huge numbers of client threads they can't reduce. =Rob
