still are some UNIX clients around whose NFS v3 implementations leave
something
to be desired on the stability and performance fronts.
The question was about TCP, not the version of the NFS protocol.
Sorry. That should have read "NFS v3 implementations and/or NFS vX on TCP implementations".
The reason for that choice had nothing to do with stability nor with any client issues so far as I know, but rather because most customers see a performance hit when running TCP with enabled,
A performance hit which is oft blamed, sometimes deservedly sometimes not, on the quality of the client implementation. Arguably the mount(1M) command on the client side should favor the higher-performing-under-most-circumstances option (UDP), but NetApp doesn't control that so...
Keith
http://www.geek-girl.com/bugtraq/1999_1/0594.html
I think that falls into the "Doc, it hurts when I do that" category, but it's a good reminder anyway.