[Xitami] 50+ simultaneous users...

[Paul Reid]

>>What O/S?

> Win2000 ...  Hope this helps!

Doesn't help me, I just thought some wiser person would know the
limits. You later suggested 128 processes which is in the
ballpark  with your observed 100 process choke-point. I don't
know the limits, but since much NT core comes from OS/2 which is
pushing 2 decades old, but MS has had fingers in it, those
numbers are not implausible. 

Just tossing off dumb ideas....

How many files are open when you choke? Aside from your program
and data files, there may be temporary files. NT seems to handle
large numbers of live files, but I've never known the limits. 

Can you test with Xitami as a non-Service process? I don't know
if that changes anything.

Can you test on a unix platform known to have larger (or at least
very different) limits on processes? Even if your spawn processes
are not available on unix, any non-trivial filter program using
similar resources (processes, file-handles) ought to suggest if
it is the O/S or Xitami.

Can you test under Apache? If it is indeed something in Xitami,
Apache should choke differently.

 -PRR