I'm running a FreeBSD 5.3-p5 server and several FreeBSD 4.11 clients.
The clients run high levels of concurrency (web servers running several
hundred processes at a time). The clients NFS connection tend to lockup
when running nfsiod but (so far) appear not to when not running nfsiod.
When the lockup occurs the send-q and recv-q on both ends tends to have
somewhere around 33000 bytes. Even when the sendspace and recvspace is
set to 65536. Any idea what each side is waiting for?

If there's something I can do to help debug the issue let me know.

--
Michael Conlen