Win9x and TNG status?

[Luke Kenneth Casson Leighton]

> > if anyone else wants to try this, i'd really appreciate some feedback.
> 
> Well, if I get TNG going tonight, I'll run it at a TNG PDC.

cool.  i'll be around, if you have questions.

> > 600 of the requested 10,000 logins.  actually, i think it was 
> > 100,000 that
> > i requested, only 500-600 succeeded.
> 
> Uhm, what kind of hardware?  Hopefully I can borrow an P166 type machine
> from work for an NT PDC, but the TNG server will be a dual-proc SPARC with
> 128MB of ram, running almost nothing else.  

i was using a p166 or a p200 with 128mb ram.
 
> > 
> > TNG fares a little better, except that i came across a bug in tdb that
> > causes it to fail, and _stay_ failed.  if anyone can repro 
> > this, please
> > SAVE your netlogoncreds.tdb file because we REALLY need to track this
> > down.
> 
> Again, if I can get this running (my incompetence is the barrier here :),
> I'll leave it running tonight and all day tomorrow, and we'll see how fast
> it brings that machine to it's knees.

check how many processes it can handle, you need one smbd, one netlogond
and one lsarpcd per user (not all at the same time, they are used and then
diee pretty quick).

there's no logging of the number of successful or failed logins, so if you
output stdout to a file and count the number of "yeses" and "nos" with
some scripts...


also, remember rpctorture forks() too so if you want 100 connections,
that's 100 rpctortures, 100 smbds, about 50-80 netlogonds and 50-80
lsarpcds (depends on who's up and who's down, and when :)

doesn't apache keep a few daemons "on ice", on the select() loop?