Subject: Re: overcommit
To: None <tech-kern@netbsd.org>
From: Chris G. Demetriou <cgd@netbsd.org>
List: tech-kern
Date: 03/20/2000 13:10:55
cgd@netbsd.org (Chris G. Demetriou) writes:
> Michael Richardson <mcr@solidum.com> writes:
> > For the curious, linux-kernel is having a debate on overcommit.
> >
> > http://www.uwsg.indiana.edu/hypermail/linux/kernel/0003.2/0003.html
>
> *chuckle* Of course, we all individually know what the one true right
> answer is... (but damned if we could agree as a group... 8-)
This statement, of course, has prompted several people (so far 8-) to
ask me what The One True Right Answer (or at least, my answer) is.
Anyway, the way i put it to the first person requesting it:
> (1) do proper accounting of pages in use and pages committed.
>
> (2) provide a switch, via sysctl, that turns on/off overcommit. allow
> root to always enable overcommit, allow root to disable overcommit
> whenever not currently overcommitted.
>
> default to one of the above, i'm not sure i care which. 8-) (I can
> see reason for overcommit in e.g. install kernels, where you have no
> swap, but think that in normal use you may well want it disabled by
> default, so people adminning servers have just one fewer knob to set
> to get the system into the 'right' configuration.)
cgd
--
Chris Demetriou - cgd@netbsd.org - http://www.netbsd.org/People/Pages/cgd.html
Disclaimer: Not speaking for NetBSD, just expressing my own opinion.