Short Roundup of Question: Kernel Tables filling up, computed new Params
as per the Docs, obtained Sizes that looked quite high, asked the List
whether they're reasonable. Complete Question below.
Answer: The Formulas used to compute the new Parameters are a Bit out-
dated and do in Fact give quite high Values. What I finally did was to
change MAXUPRC to (NPROC/3) (everyone should do this unless they have
a very small Process Table!) and up maxusers to 64 (previously 16, For-
mulas gave 128). Bingo! File and Process Table Usage dropped from
50-70% to <20%; during light to medium Usage, that is. No more File
Table going boom so far. ;-)
Thanks to:
gdonl@gv.ssi1.com (Don Lewis)
(Suggestion: maxusers 32, up derived Params as needed)
Kevin.Sheehan@uniq.com.au (Kevin Sheehan)
(Up Params individually, or even better, up OS to 2.5)
Glenn.Satchell@uniq.com.au (Glenn Satchell)
(maxusers 64)
Original Question:
> Just a quick Question: I received Word today that it might be Time to
> optimize our Kernel Adjustments. (Read: File Table full. ;-) I'll start
> out with the main User-accessible Machine; That's SunOS 4.1.3_U1B, sun4m
> (SS10/71), 128 MB RAM, differential SCSI with small Disk Farm. I'll lower
> MAXUPRC (relatively to NPROC) while I'm at it.
>
> The "Problem" is that I typically have 10-20 Users, 60-100 Windows (inclu-
> ding non-Shell Windows and Windows displaying to another Display) and
> 30-40 NFS Clients. According to the "Book", this gives me maxusers=128,
> NPROC=2058, MAXUPRC (now NPROC/3) =686 and nfile=7110, which sounds a
> Bit far out of the Window compared to 16/266/261/1017 now. Is this
> reasonable, or am I going to bury my RAM under a Monster Kernel? :-}
Thanks again,
J. Bern
-- /\ /""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""\ / \/ bern@uni-trier.de (Size Limit!) | P.O. Box 1203 | Ham: \/\ / J. \ bern@ti.uni-trier.de (SUNAttachm.OK) | D-54202 Trier | DD0KZ / \ \Bern/ No Finger etc.; Use Mail (Subj. "##" for Autoreply List) and \ / \ /\ WWW. /\/ \/ \____________________________________________________________/
This archive was generated by hypermail 2.1.2 : Fri Sep 28 2001 - 23:11:00 CDT