>I've currently got my filers with 'options nfs.tcp.enable off' from when I
first
>set them up due to a perception that this would cause problems.
>
>Has anyone got tcp enabled? Have they seen any problems/benefits, or are
there
>any known issues with this?
I think the "problem" with switching on nfs.tcp.enable stems from the way that
the mount(1M) command works on many modern UNIX systems. If you don't actually
specify the version of NFS that you want to use, and the transport protocol you
wish it to ride on the mount(1M) command line, the default is often to use the
highest version of NFS available (usually 3) over the first connection oriented
transport found (usually TCP), that's supported by both the client *and* the
server. So, if you're a little "lazy" when it comes to telling the system
exactly what you want, you'll end up with NFS v3 on TCP when what you
*probably* wanted was UDP and/or NFS v2.
Although NFS v3 over TCP or UDP is "fine by us" at the server end of the show
(NFS is one of the small number of things a filer does, so you'll find it's
done rather well no matter what flavour you pick ;-)), there have been and
still are some UNIX clients around whose NFS v3 implementations leave something
to be desired on the stability and performance fronts. Hence the default
disablement of some of the fancier combinations, to prevent them from being
selected by accident.
Keith