Buffer space and socket syscall tracing

[Andrew Pantyukhin]

I've stumbled upon the infamous "No buffer space
available" problem (with thousands of netgraph
nodes). I'm wondering if there's a really comprehensive
way to debug it. I know that it's probably a matter
of some sysctl, but I've googled and tuned for days -
and I now feel that there must be another way.

Truss and ktrace both say that the error is generated
when a socket syscall is made. Is there a way to
trace inside the syscall and find the very particular
reason for this? I know that it might not be of any use
to me as I'm not any kind of kernel hacker, but I
want to give it a shot.

Thanks