Brian's problem...

[Brian K. White]

----- Original Message ----- 
From: "John Esak" <john at valar.com>
To: "Fplist (E-mail)" <filepro-list at seaslug.org>
Sent: Thursday, November 25, 2004 11:38 AM
Subject: Brian's problem...


>
> Sorry, deleted the thread and subject.
>
> But, I meant to mention in my last post that MAXUP is probably the 
> resource
> Ken was referring to as the one which limits user processes. It used to be
> 50 (hmmmm) on older SCO systems, but is now 100 as a default.  This once 
> bit
> me pretty bad. I kept having all sorts of failures.... because I didn't 
> know
> that they hired a new user (this was back in about 1987 or so) and they 
> just
> let her use the same login id as another girl. The _strangest_
> undiscoverable problems were happening, until I tracked it down to MAXUP
> going over 50!  Then they told me that the new "mary" was logging in with
> the old mary's login becuase they didn't know how to add a new user. I 
> fixed
> the problem first by giving the new mary the name maryd and later just
> changed the tunable to 100.
>
> Totally off subject but the same kind of inscruatable errors will bite you
> if FLCRECS or (maybe its NUMFLKRECS or whatever) is too small. Strnge 
> memory
> region hassles start happening and they expose themselves as completely
> strange failures at completely odd places... with hardly any way to trace
> them.
>
> John

Thanks for the leads John.
MAXUP was already 100, Bela says the limit is 11,000 since 4.2, I upped it 
to 255
KERNEL_FLCKREC was 256
KERNEL_FLCKREC_MAX is 16000
so I upped KERNEL_FLCKREC to 512

...reboot...  it acts exactly the same. right down to the wierd ness of ksh 
working at levels 1 through 28, 43, and 45

I like those changes though for this system so it wasn't a waste exactly, it 
just had no effect on this problem.

I really need to get sarcheck or olympus on this box though, because didling 
these values it's easy to do something that sounds good "allow lots of X" 
when really you are doing something bad "no Y available because the space 
was all given to X"

Brian K. White  --  brian at aljex.com  --  http://www.aljex.com/bkw/
+++++[>+++[>+++++>+++++++<<-]<-]>>+.>.+++++.+++++++.-.[>+<---]>++.
filePro BBx  Linux SCO  Prosper/FACTS AutoCAD  #callahans Satriani