On Thu, Nov 16, 2017 at 6:54 AM, Frank Filz <[email protected]> wrote:
>> For me, its the simplicity of it, and the legacy of it working for a long >> time, and >> the utter lack of modern best practices documentation. >> If you go looking for NFS howto's, you almost immediately notice that they >> are all at least 10 years old, and there is pretty much nowhere to ask >> questions. At least, that has been the case in the past when I've tried to >> go >> looking. UDP in particular is stateless, which is a nice feature. > > It's a bit surprising there isn't much NFS how to online, but most of the > NFS developer community is involved with enterprise servers and probably the > how to stuff is in product installation guides and such... > > If you don't really require fcntl locks, then NFS v3 is certainly a simple > protocol for transferring data, though as I mentioned before, using UDP is > now strongly discouraged due to the fragment reassembly issues causing both > severe performance issues and data integrity issues. > > NFS v4 doesn't always perform as well as NFS v3, but if you are doing fcntl > locking, it's far superior for resiliency across server and client crashes. > > I can try and answer questions here... I know a thing or two about NFS > (having worked with both the Linux kernel NFS implementation and the > nfs-ganesha user space server)... Oh, are you the fellow who gave the ganesha talk a few years ago? For me, NFSv3 UDP has been mostly just working for me. The one thing that has caused me grief is GIMP, which seems to have egregiously bad save performance on my NFS mounts. _______________________________________________ PLUG mailing list [email protected] http://lists.pdxlinux.org/mailman/listinfo/plug
