Path: utzoo!utgpu!jarvis.csri.toronto.edu!mailrus!wuarchive!gem.mps.ohio-state.edu!tut.cis.ohio-state.edu!bloom-beacon!spdcc!merk!alliant!linus!nixbur!nixpbe!net4!netmuc!bothe
From: bothe@netmuc.UUCP
Newsgroups: comp.sys.pyramid
Subject: trouble changing nfs_sendspace in p
Message-ID: <10000010@netmuc>
Date: 13 Oct 89 05:39:00 GMT
Lines: 33
Nf-ID: #N:netmuc:10000010:000:974
Nf-From: netmuc.UUCP!bothe    Oct 13 06:39:00 1989


Targon 35 Release TOS 3.3.03

I tried to optimize the system by increasing the amount of space that is
reserved for the nfs port socket send and receive buffers.
(We have heavy nfs traffic!)
I changed in /usr/sys/conf/param.c
nfs_sendspace from 27000 to 270000
nfs_recvspace from 63000 to 630000

The result was that the nfsd (/etc/nfsd 8 in /etc/rc.local) died
immediately without saying anything.

Is this behavior correct and why does no warning occure in param.c ?
Is it a good idea increasing these values and what are the limits ?
Is there a course, literatur or anything to learn optimizing the
system handling number or size of systembuffers ?

|\       |    |
| )      |    |
|<   _  _|_   |/~\  /~\
| ) | |  |    |  | (___)
|/  |_|   \_  |  |  \__

Richard Bothe
NIXDORF COMPUTER AG
Dep DTSW 2-4
Berliner Str. 95
D-8000 Munic (West Germany)
Tel.: (nat-89)3601-2956
eunet: bothe.muc@nixpbe.UUCP
!USA: mcvax!unido!nixpbe!bothe.muc
USA:  uunet!linus!nixbur!bothe.muc