On Nov 7, 2014, at 6:57 AM, Stefan Fuhrmann <stefan.fuhrm...@wandisco.com> wrote: > > > My side of the bargain has been trying to think of unnecessary > restrictions that FSFSv7 might have and then to fix them. It > started with performance in "unmanaged" setups, continued > with support for heterogeneous clusters and currently ended > with defined behaviour after hot upgrade. I also "shot bullets" > at repositories to see whether random corruption could sneak > past our consistency checks. And I will continue testing more > and more scenarios. > > -- Stefan^2. >
I just want to say that while my personal belief is that we should largely leave fsfs alone, I do appreciate from an engineering perspective the amount of effort you have put in to trying to make this code bullet proof and also to measure the improvements. I think you have done about as much and as well as one person can. Mark
