On Sat, Apr 16, 2011 at 12:00 AM, Anthony Liguori <aligu...@us.ibm.com> wrote: > 3) We have no way to detect when we no longer need a work around which makes > (2) really unappealing.
I agree. > 4) That leaves us with: > a) waiting for NFS to get fixed properly and just living with worse > performance on older kernels > > b) having a user-tunable switch to enable bouncing > > I really dislike the idea of (b) because we're stuck with it forever and > it's yet another switch for people to mistakenly depend on. The user-tunable switch is potentially interesting for performance troubleshooting. We have seen another file system which has issues with vectored direct I/O. It would have been much easier to identify the problem by telling the user "Try running it with linearize=on and see if it makes a difference". But let's try harder on linux-nfs. Stefan
