Re: Server patch levels...

• Date: Wed, 22 May 1996 20:22:01 PDT
• From:
  Pavel Curtis <pavel@parc.xerox.com>
• In-reply-to: Christopher Scott Delaney's message of "Wed, 22 May 1996 16:53:19 PDT" <Pine.OSF.3.91.960522184410.14247A-100000@copper.ucs.indiana.edu>

Christopher Scott Delaney writes:
> We've changed NOTHING on the system.  The types and number of tasks 
> running on the machine are the SAME.  The MOO's background tasks are the 
> same.  We've ruled out everything except the moo server.  

You said *that* before, and then you come back later and point out that you
changed the FUP code as well...  I suggest that you look more carefully at the
extensions you're adding into the server; I do *not* guarantee in my releases
that any of these extensions will continue to work.  Perhaps you have other
extensions, aside from FUP?

Very little has changed in the server between p2 and p5, and I have reviewed
every single change to see if I could *imagine* it causing your lag; no dice.
Sorry.

> It is disconcerting that it looks like we have two choices: Grin and smile
> and hope it just goes away or lose a LOT of work in order to convince people
> it is indeed p4-5 that has caused this increase in CPU useage and thus
> extremely sluggish server.

I understand your frustration, but there really isn't anything more I can do
from here.  Since *I* can't do anything for you, it really does seem like *you*
are going to have to.  One option you might try, in the same model as I
suggested before, with the side-by-side comparisons, is to get a knowledgeable
C programmer to hack up the server Makefile a bit to create a `profiling'
version of the server (i.e., compiling with the `-p' option on).  You could
then do this to both p2 and p5, run the same DB on both, and compare the
`gprof' outputs from them to try to pinpoint the part of the server that's
running more slowly (if any).

	Pavel

• Re: Server patch levels...
• From: Christopher Scott Delaney <cdelaney@copper.ucs.indiana.edu>

• Prev: Re: Server patch levels...
• Next: set_connection_option()
• Index: MOO-cows
• Thread: MOO-cows