Xref: utzoo comp.sys.dec:3736 comp.unix.ultrix:4203
Path: utzoo!utgpu!news-server.csri.toronto.edu!cs.utexas.edu!uunet!cbmvax!grr
From: grr@cbmvax.commodore.com (George Robbins)
Newsgroups: comp.sys.dec,comp.unix.ultrix
Subject: Re: Swap space question...
Message-ID: <13707@cbmvax.commodore.com>
Date: 8 Aug 90 18:51:44 GMT
References: <1990Aug8.165421.27886@wam.umd.edu>
Reply-To: grr@cbmvax (George Robbins)
Organization: Commodore, West Chester, PA
Lines: 39

In article <1990Aug8.165421.27886@wam.umd.edu> rgc@wam.umd.edu (Ross Garrett Cutler) writes:
>  I recently did a generic (default) setup of Ultrix 3.1d on a DS3100 w/8 MB
>RAM and an RZ55 (300 MB HD).  I just did a pstat -s, and it said that 64 MB was
>allocated for swap space!  Now I always thought that you should allocate
>about double the installed RAM, but not 8 times!  Note that ~11 MB was
>currently being used in an idle 1-user state (most likely because of that
>RAM-hog DEC Windows).  Anyway, are there any recommendations as to how
>much swap we should allocate?

With windowing on the RISC based systems, you should allocate lots of swap
space - I'd say at least 4X for a personal system, more if you get into
multi-tasking, emacs or applications abuse.
 
Actual allocation is probably derived from standard disk partitions, rather
than any configuration dependent "need", so you might want to tune it a bit.

>On a similar front, we also have a VS3200 running Ultrix 3.0 w/8 MB RAM.
>We have 16 MB of swap space allocated; if we try to run two programs that
>should not fill up the swap space (each uses about 1 MB of text and data
>combined), we get messages like "out of core".  Has anyone else had
>this problem?  And how much swap space (in general) is recommonded for this
>system?

You need to take a look at all the swap space requirements for all the active
processes.  Ultrix allocates swap space for each process, whether or not it
needs it, so you may find the sum taken up by extra logins, daemons, etc to
be more than you expect. 

The units of allocation are controlled by the dmmin and dmmax config parameters.
For swap constrained systems, you want to make sure that dmmin is large  enough
that getty and the like fit within the iniital allocation, elsewise they start
chewing into the dmmax units...

There may also be some fragmentation  issues, but I haven't dug into it that
deeply.
-- 
George Robbins - now working for,     uucp:   {uunet|pyramid|rutgers}!cbmvax!grr
but no way officially representing:   domain: grr@cbmvax.commodore.com
Commodore, Engineering Department     phone:  215-431-9349 (only by moonlite)