Skip site navigation (1)Skip section navigation (2)
Date:      Sun, 6 Apr 2003 23:47:54 -0700 (PDT)
From:      Doug White <dwhite@gumbysoft.com>
To:        Mattias Pantzare <pantzer@ludd.luth.se>
Cc:        freebsd-stable@freebsd.org
Subject:   Re: load testing and tuning a 4GB RAM server
Message-ID:  <20030406234549.C54715@carver.gumbysoft.com>
In-Reply-To: <3E90835F.9020104@ludd.luth.se>
References:  <20030406120112.I11095-100000@netmint.com> <3E906828.8060001@mac.com> <3E90835F.9020104@ludd.luth.se>

next in thread | previous in thread | raw e-mail | index | archive | help
On Sun, 6 Apr 2003, Mattias Pantzare wrote:

>
> > The notion of tuning kernel variables to the amount of memory on the
> > _build_ system-- which is not always the same as the system which will
> > be running the kernel-- doesn't strike me as the right thing to do, so I
> > tend to set this to 64 or so by default, and then change it on a
> > case-by-case basis as the need arises.
>
> That is why maxusers 0 will auto-size when the kernel _boots_. Not at
> buildtime.

.. unfortunately 'maxusers 0' on >2GB systems causes the tables to get
*too* big, and you can easily deplete KVM in this instance. Check the
archives for the discussion-of-the-month about this.

I'd suggest maxusers 128 and tune other items as needed, particuarly the
tunables kern.ipc.nmbclusters and kern.ipc.nmbufs.

-- 
Doug White                    |  FreeBSD: The Power to Serve
dwhite@gumbysoft.com          |  www.FreeBSD.org



Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?20030406234549.C54715>