cvs commit: src/sys/i386/include vmparam.h

[David O'Brien]

On Mon, Aug 16, 2004 at 07:42:21PM -0600, Scott Long wrote:
> I want to know why the limit that you chose is attributed to me and a
> quick suggestion that I made in an email?

You gave me the figure of 100000 for kern.maxvnodes max limit.  You
sounded rather authoritative at the time, that a general purpose
workstation wouldn't want more than that.  You've so back peddled from
this I'm sorry I took your suggested value.  RU has complained I wasn't
giving 0 attributes in commit messages, so I gave you attribution.  I
won't make that mistake again.

So you're saying that a kern.maxvnodes of 150000 is a good default figure
on a 4GB machine?  If it bothers you that much I'll bump MAXVNODES_MAX or
remove the limiting.  I personally haven't needed kern.maxvnodes this
high on a general purpose workstation or department server.

 
> I'll say it again.  It's a hard problem.  If I knew the easy answer I
> would have committed it long ago.

Like everything else in FreeBSD'ville -- one works on the itch they are
experiencing at the moment.  I'm scratching the 'panicing on lightly
loaded 4GB systems isn't acceptable' itch since I'm itching all over.

-- 
-- David  (obrien at FreeBSD.org)